{ "site": "AI-nterprise", "language": "en", "base_url": "https://ai-nterprise.com", "preview": false, "public_release_authorized": true, "book_status": "Forthcoming · Q4 2026 · English edition", "snapshot_sha256": "53cfcf63a09b978ed618a84ef36416da2560f056134c7ed748505a99f8c1249d", "schema_version": "1.0", "pages": [ { "path": "/about/", "title": "About Ottaviano De Cicco", "description": "Ottaviano De Cicco, technology executive and advisor, and author of AI-nterprise. Enterprise technology, digital transformation and AI governance.", "content_status": "published", "h1": "Ottaviano De Cicco", "canonical": "https://ai-nterprise.com/about/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/about.md", "html_sha256": "81a89989ea7c4f3cc08b2b7a3ed6c0ea930604f4608f1680d557664783cd7b5d", "markdown_sha256": "e9445c074a62d08948ce00bf3a18709b371977c4f54d07cd5095f6a25960ff45", "content_markdown": "Canonical: https://ai-nterprise.com/about/\nLanguage: en\nContent status: published\n\nAbout · The author\n\n# Ottaviano De Cicco\n\nI am a technology executive and advisor with over 20 years of commercial leadership experience across enterprise technology, digital transformation and cybersecurity. My current work focuses on how organizations adopt AI and govern its use.\n\nIn AI-nterprise, I explore how organizations govern knowledge, judgment and authority as people and AI agents work together. The book and open framework offer a practical vocabulary for leaders facing these decisions.\n\n[Professional background on LinkedIn ↗](https://www.linkedin.com/in/ottavianodecicco)\n\n## Books & papers\n\n- [AI-nterprise: Governing What the Organization Knows · White paper, v1.0 · 3 September 2026 ↗](https://doi.org/10.5281/zenodo.22282125)\n\n- [WebSigil: A Measurement Authority Architecture for AI Reliability · Technical report, v1.0 · 3 July 2026 ↗](https://doi.org/10.5281/zenodo.21177726)\n\n- [AI-nterprise: A Manifesto for the Human-Agent Enterprise · 4 September 2026](https://ai-nterprise.com/articles/manifesto/)\n\n- [AI-nterprise · Forthcoming · Q4 2026 · English edition](https://ai-nterprise.com/book/)\n" }, { "path": "/action-boundary/", "title": "An illustrative action boundary", "description": "A synthetic refund proposal tests an active mandate and current claim version. No payment is executed.", "content_status": "published", "h1": "A recommendation meets its boundary.", "canonical": "https://ai-nterprise.com/action-boundary/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/action-boundary.md", "html_sha256": "998d08b46c1f8abc283b9dc89e05024e802fad77d11d4deaff1b40843d651e8c", "markdown_sha256": "eb64664701838c032cd9d371f942caba834b142265021c2c0883427d15a90080", "content_markdown": "Canonical: https://ai-nterprise.com/action-boundary/\nLanguage: en\nContent status: published\n\nAn illustration / Synthetic data\n\n# A recommendation meets *its boundary.*\n\nA €240 refund proposal. Change one condition and inspect what happens at the point of action.\n\nIllustrative browser demonstration. No payment, external system or product is connected.\n\nRuns in your browser. Nothing is sent or stored by us.\n\n01 / Change the conditions\n\n## What is true at the moment of use?\n\n**Mandate at time of action**\n\n- Active · refund up to €300\n\n- Expired · no current authority\n\n**Current claim version**\n\n- v1 · same as the proposal\n\n- v2 · corrected after the proposal\n\nEvaluate the proposal\n\nEnable JavaScript to run this local illustration. Its complete rule is explained below.\n\n02 / Decision record\n\n## Ready to evaluate\n\nThe proposal has not been checked. No refund will be executed in this example.\n\n**Proposal**\n\nRefund €240 · claim v1\n\n**Checks**\n\nNot yet evaluated\n\n**Action**\n\nSimulation only · Nothing executed\n\n### Inspect the same record as JSON\n\n```\n{\"status\": \"not_evaluated\", \"action_executed\": false}\n```\n\nChapter 9 / Four Decisions Hiding Inside “Approved”\n\n## A proposal is eligible only within these conditions.\n\nThis illustration relates to Chapter 9 of AI-nterprise, which separates relying on a claim from authorizing an action.\n\nA proposal is eligible in this illustration only if the mandate is active, the amount is within its limit, and the referenced claim version is current. Changing a selection clears the previous result.\n\nA corrected claim does not establish that a refund should be refused. It requires a new assessment using the current version.\n\nThe limit\n\n## What remains outside the example.\n\nClaim truth, identity verification, authorized claim updates, distributed enforcement, concurrent revocation and repair of earlier actions. Production systems must address these separately.\n\n### Data in this illustration\n\nThe controls change this synthetic example locally. The selections are not submitted to a service or saved between visits.\n\n[Explore the reference model](https://ai-nterprise.com/framework/)\n" }, { "path": "/agents/", "title": "For AI agents", "description": "Read the current AI-nterprise website as HTML, per-page Markdown, full text or structured JSON, with canonical sources and version hashes.", "content_status": "published", "h1": "For AI agents", "canonical": "https://ai-nterprise.com/agents/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/agents.md", "html_sha256": "6b05e6382a0702a2988e116f104cd889b96c1360c1461256af936dea875c7aa0", "markdown_sha256": "75d44579ff10dc27fa8a6f94661fdccb8fd6596ece2c8f03c25a53a6c028da73", "content_markdown": "Canonical: https://ai-nterprise.com/agents/\nLanguage: en\nContent status: published\n\nPublic reading formats · English\n\n# For AI agents\n\nThe same public content.\nReady to read, retrieve and cite.\n\n## One source. Several reading formats.\n\nThese formats are generated from the pages served on this site. They preserve the public wording, canonical links, source attributions and text descriptions of the framework figures. No account or JavaScript is needed to read them.\n\n**The book:** Forthcoming · Q4 2026 · English edition\n\nStart here · Text\n\n### [A compact site guide](https://ai-nterprise.com/llms.txt)\n\nPage descriptions, canonical links, reading formats and original publications.\n\n/llms.txt\n\nRead everything · Text\n\n### [All public page text](https://ai-nterprise.com/llms-full.txt)\n\nThe complete text of the public pages in one document, with links and diagram transcripts.\n\n/llms-full.txt\n\nBrowse · JSON\n\n### [A structured index](https://ai-nterprise.com/agent-index.json)\n\nPage metadata, source references, Markdown locations and content hashes.\n\n/agent-index.json\n\nRetrieve · JSON\n\n### [Full page content](https://ai-nterprise.com/agent-content.json)\n\nThe same Markdown text, grouped by canonical page and tied to the served HTML by a hash.\n\n/agent-content.json\n\n## Read by page\n\nChoose the original HTML or its generated Markdown representation.\n\n- ### [Knowledge, judgment, and authority](https://ai-nterprise.com/)\n\nThe book, open framework and practical resources for the human–agent enterprise, by Ottaviano De Cicco.\n\n[Read the page](https://ai-nterprise.com/) [Read Markdown](https://ai-nterprise.com/agents/content/home.md)\n\n- ### [About Ottaviano De Cicco](https://ai-nterprise.com/about/)\n\nOttaviano De Cicco, technology executive and advisor, and author of AI-nterprise. Enterprise technology, digital transformation and AI governance.\n\n[Read the page](https://ai-nterprise.com/about/) [Read Markdown](https://ai-nterprise.com/agents/content/about.md)\n\n- ### [An illustrative action boundary](https://ai-nterprise.com/action-boundary/)\n\nA synthetic refund proposal tests an active mandate and current claim version. No payment is executed.\n\n[Read the page](https://ai-nterprise.com/action-boundary/) [Read Markdown](https://ai-nterprise.com/agents/content/action-boundary.md)\n\n- ### [Ideas](https://ai-nterprise.com/articles/)\n\nRead the full manifesto and essays by Ottaviano De Cicco on institutional design, organizational learning, and human judgment.\n\n[Read the page](https://ai-nterprise.com/articles/) [Read Markdown](https://ai-nterprise.com/agents/content/articles.md)\n\n- ### [AI-nterprise: A Manifesto for the Human-Agent Enterprise](https://ai-nterprise.com/articles/manifesto/)\n\nThe institutional question behind the AI-nterprise framework.\n\n[Read the page](https://ai-nterprise.com/articles/manifesto/) [Read Markdown](https://ai-nterprise.com/agents/content/articles/manifesto.md)\n\n- ### [The Human Hallucination: The AI Failure Mode Nobody Is Measuring](https://ai-nterprise.com/articles/the-human-hallucination/)\n\nWhy human judgment and confidence must be assessed alongside machine output.\n\n[Read the page](https://ai-nterprise.com/articles/the-human-hallucination/) [Read Markdown](https://ai-nterprise.com/agents/content/articles/the-human-hallucination.md)\n\n- ### [The Model Is Rented. The Learning Must Be Yours.](https://ai-nterprise.com/articles/the-model-is-rented/)\n\nHow an organization retains what it learns while using rented models.\n\n[Read the page](https://ai-nterprise.com/articles/the-model-is-rented/) [Read Markdown](https://ai-nterprise.com/agents/content/articles/the-model-is-rented.md)\n\n- ### [The book](https://ai-nterprise.com/book/)\n\nAI-nterprise by Ottaviano De Cicco: the argument, open white paper and free printable resources.\n\n[Read the page](https://ai-nterprise.com/book/) [Read Markdown](https://ai-nterprise.com/agents/content/book.md)\n\n- ### [Contact Ottaviano De Cicco](https://ai-nterprise.com/contact/)\n\nContact Ottaviano De Cicco about AI-nterprise and the governance of enterprise AI workflows.\n\n[Read the page](https://ai-nterprise.com/contact/) [Read Markdown](https://ai-nterprise.com/agents/content/contact.md)\n\n- ### [Framework · A visual reference model](https://ai-nterprise.com/framework/)\n\nFollow the AI-nterprise model visually: four decisions, three control planes and three coupled loops.\n\n[Read the page](https://ai-nterprise.com/framework/) [Read Markdown](https://ai-nterprise.com/agents/content/framework.md)\n\n- ### [Vocabulary](https://ai-nterprise.com/glossary/)\n\nDefinitions and exact formulations from the AI-nterprise framework, with direct term links and the source DOI.\n\n[Read the page](https://ai-nterprise.com/glossary/) [Read Markdown](https://ai-nterprise.com/agents/content/glossary.md)\n\n- ### [Resources](https://ai-nterprise.com/tools/)\n\nDownload Four Questions Before an Agent’s Answer Becomes Action: instructions, a process map, a filled example, and an open licence.\n\n[Read the page](https://ai-nterprise.com/tools/) [Read Markdown](https://ai-nterprise.com/agents/content/tools.md)\n\n## Original publications\n\nPublication dates below belong to the original works. A content hash identifies a site snapshot; it is not a publication date.\n\n- [AI-nterprise: A Manifesto for the Human-Agent Enterprise ↗](https://www.linkedin.com/pulse/ai-nterprise-manifesto-human-agent-enterprise-ottaviano-de-cicco-jjmbe) 2026-09-04\n\n- [AI-nterprise: Governing What the Organization Knows ↗](https://doi.org/10.5281/zenodo.22282125) 2026-09-03 · v1.0\n\n- [The Model Is Rented. The Learning Must Be Yours. ↗](https://www.linkedin.com/pulse/model-rented-learning-must-yours-ottaviano-de-cicco-16xsf) 2026-07-15\n\n- [WebSigil: A Measurement Authority Architecture for AI Reliability ↗](https://doi.org/10.5281/zenodo.21177726) 2026-07-03 · v1.0\n\n- [The Human Hallucination: The AI Failure Mode Nobody Is Measuring ↗](https://www.linkedin.com/pulse/human-hallucination-ai-failure-mode-nobody-measuring-de-cicco-exjhf) 2026-06-12\n\n## Keep the context attached.\n\nCite a page by its canonical URL and retain the author, source, version and limits stated there. The original publications remain the references for their respective claims. The framework figures have text transcripts; motion and layout are available on the HTML pages.\n\nThe action-boundary demonstration and film are illustrations. They do not establish product capabilities or enterprise outcomes. These endpoints provide public reading material, not an API for executing actions.\n\n## Scope and reuse\n\nThe exports contain the public website. They do not include the unpublished book manuscript, private drafts, subscriber information or project files.\n\nA reading format does not change the licence of a work. The worksheet and white paper retain the licences stated on their source pages; other author texts retain their existing copyright and attribution.\n\n[Contact the author](https://ai-nterprise.com/contact/)\n" }, { "path": "/articles/", "title": "Ideas", "description": "Read the full manifesto and essays by Ottaviano De Cicco on institutional design, organizational learning, and human judgment.", "content_status": "published", "h1": "Read beyond the answer.", "canonical": "https://ai-nterprise.com/articles/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/articles.md", "html_sha256": "c270176eae9db3e61455f2d6653c82942ce709342d03f06b0fee5271a3eb783a", "markdown_sha256": "d2680d8f11745b008eeb4242064580d85f04ef24a41dc2008a6dc5d2a034802c", "content_markdown": "Canonical: https://ai-nterprise.com/articles/\nLanguage: en\nContent status: published\n\nIdeas · The open argument\n\n# Read beyond the answer.\n\nExplore institutional design, organizational learning, and human judgment through the author’s original publications.\n\n![](https://ai-nterprise.com/assets/editorial-manifesto.webp)\n\nFoundation 4 September 2026 11 min read\n\n## [AI-nterprise: A Manifesto for the Human-Agent Enterprise](https://ai-nterprise.com/articles/manifesto/)\n\nThe institutional question behind the AI-nterprise framework.\n\n[Read the manifesto](https://ai-nterprise.com/articles/manifesto/)\n\n![](https://ai-nterprise.com/assets/editorial-learning.webp)\n\nLearning 15 July 2026 11 min read\n\n## [The Model Is Rented. The Learning Must Be Yours.](https://ai-nterprise.com/articles/the-model-is-rented/)\n\nHow an organization retains what it learns while using rented models.\n\n[Read the essay](https://ai-nterprise.com/articles/the-model-is-rented/)\n\n![](https://ai-nterprise.com/assets/editorial-judgment.webp)\n\nJudgment 12 June 2026 10 min read\n\n## [The Human Hallucination: The AI Failure Mode Nobody Is Measuring](https://ai-nterprise.com/articles/the-human-hallucination/)\n\nWhy human judgment and confidence must be assessed alongside machine output.\n\n[Read the essay](https://ai-nterprise.com/articles/the-human-hallucination/)\n" }, { "path": "/articles/manifesto/", "title": "AI-nterprise: A Manifesto for the Human-Agent Enterprise", "description": "The institutional question behind the AI-nterprise framework.", "content_status": "published", "h1": "AI-nterprise: A Manifesto for the Human-Agent Enterprise", "canonical": "https://ai-nterprise.com/articles/manifesto/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/articles/manifesto.md", "html_sha256": "586f88875c2f65baa5f5c6d7e4358c5ccff4124d4e598778fa98dd26acc4946a", "markdown_sha256": "7942b41d9a989c5b8b9b0930c034e6f21c841a38fb694b1dbd40210d9a247258", "content_markdown": "Canonical: https://ai-nterprise.com/articles/manifesto/\nLanguage: en\nContent status: published\n\n[Ideas](https://ai-nterprise.com/articles/) · Foundation\n\n# AI-nterprise: A Manifesto for the Human-Agent Enterprise\n\nOttaviano De Cicco · 4 September 2026 · 11 min read\n\nFirst published on [LinkedIn ↗](https://www.linkedin.com/pulse/ai-nterprise-manifesto-human-agent-enterprise-ottaviano-de-cicco-jjmbe), 4 September 2026.\n\n![](https://ai-nterprise.com/assets/editorial-manifesto.webp)\n\n*AI governance was designed to control models. The next enterprise architecture must govern what humans and agents are allowed to know, decide, remember, and do together.*\n\n*Most companies are adding AI to the enterprise. Few are redesigning the enterprise for AI.*\n\n*We are putting copilots into functions, agents into workflows, and language models between people and the information they use to decide. Yet the organization around them still assumes a familiar world: software executes instructions, people exercise judgment, records preserve what happened, and authority follows the org chart.*\n\n*That world is disappearing faster than most governance programs can adapt.*\n\n*An agent can now find evidence, interpret a policy, compare alternatives, recommend a decision, write the rationale, update a system, and leave behind a summary that another human or agent will later treat as context. It does not merely process information. It participates in the production of organizational reality.*\n\n*This is the shift that matters. The enterprise is becoming a****human-agent system****.*\n\n*Organizations are not just collections of people, software, and data. They are systems for deciding what counts as true, who may decide, what may be done, what must be remembered, and who answers when something goes wrong. AI is entering all five questions at once.*\n\n*I use****AI-nterprise****to name the institutional architecture this requires: a way to coordinate humans, agents, models, claims, and rules so that an organization can know, decide, act, and learn without allowing authority, accountability, or memory to dissolve into the workflow.*\n\n*It is not a product category. It is not a model class. And it is not another compliance checklist.*\n\n*It is a design problem for the enterprise itself.*\n\n## The enterprise has acquired a new kind of actor\n\n*Traditional enterprise software could be complex, but its role was legible. A system stored a record, applied a rule, calculated a result, or routed a task. If the business rule changed, someone changed the software.*\n\n*Agents blur that boundary. They do not only execute a predefined path. They select which evidence matters, infer what a request means, choose among tools, generate intermediate reasoning, and decide what to do next. Their behavior is partly specified in code and partly produced at runtime.*\n\n*This does not make an agent a legal person, a colleague, or an autonomous institution. But it does make it an actor in the practical sense: it can alter the premises on which other actors behave.*\n\n*That distinction is more important than the familiar question of whether AI will replace jobs. A job is a bundle of tasks. An enterprise is a distribution of authority and consequence. Once agents participate in decisions, the central question is no longer only****what can they do?****It is****what are they entitled to make true for the organization?***\n\n*An agent may be able to draft a policy. That does not mean it may establish one. It may be able to retrieve a customer record. That does not mean it may expose every representation of it. It may be able to recommend a payment, a diagnosis, a hire, or a control exception. That does not mean its recommendation has standing, or that any downstream system should act on it.*\n\n**Capability is not authority.**\n\n*The human-agent enterprise begins by refusing to confuse the two.*\n\n## Governance is looking at the wrong object\n\n*Many AI governance programs are organized around models and applications. They inventory systems, classify use cases, evaluate vendors, test outputs, document risks, and assign accountable owners. All of this is necessary.*\n\n*It is no longer sufficient.*\n\n*The object that travels through the enterprise is rarely the model. It is the****claim****the model helped produce: this customer is eligible; this contract contains a material risk; this incident has been resolved; this forecast should guide the plan; this employee should receive an exception.*\n\n*The claim moves. It enters a presentation, a ticket, a knowledge base, a CRM, a policy, a line of code, or another agent's context. Its origin becomes less visible with every handoff. A provisional answer is copied into an official document. A generated summary becomes the premise for a later decision. Several agents repeat the same unsupported conclusion and create the appearance of consensus.*\n\n*By the time the consequence arrives, asking which model produced the first sentence may be less useful than asking:*\n\n- *What exactly was asserted?*\n\n- *Which version was relied upon?*\n\n- *For what purpose was it accepted?*\n\n- *Who or what was allowed to see it?*\n\n- *What action did it enable?*\n\n- *What happened when it was corrected?*\n\n*Provenance can tell us where a claim came from. It cannot, by itself, tell us what the organization is entitled to do with it.*\n\n**Output is not knowledge.**\n\n*The difference between the two is institutional standing.*\n\n## Four decisions we must stop collapsing\n\n*A human-agent enterprise needs to make four decisions explicitly. They are related, but they are not interchangeable.*\n\n***Epistemic standing:****For which purpose, in which context, and at what level of consequence is a claim sufficiently warranted?*\n\n***Persistence:****Should the claim be retained, for how long, in which version, and under what correction or disposal rule?*\n\n***Visibility:****Which person or agent may retrieve which representation of the claim and its supporting context?*\n\n***Action enablement:****Who or what may act on the claim now, within which boundary, and with which escalation path?*\n\n*A claim may be credible without being permanent. It may be retained without being authoritative. It may be visible to an auditor but unavailable to an operational agent. It may be accepted for exploration and forbidden for execution. An actor may be permitted to act on it in one jurisdiction, function, or time window and not in another.*\n\n*Collapse these distinctions into a confidence score or an access-control list and the enterprise will make category errors at machine speed.*\n\n*Seeing is not believing. Remembering is not accepting. Accepting is not authorizing action.*\n\n*The architecture must preserve those differences even when the workflow would prefer to erase them.*\n\n## “Human in the loop” is not a governance model\n\n*The phrase sounds reassuring because it places a person somewhere in the process. It says almost nothing about whether that person still has meaningful power.*\n\n*A human can be present and still function as a ceremonial approver. If the person sees the model's conclusion before forming an independent view, the system has already framed the question, selected the evidence, and established an anchor. The interface may then record a click called “review” while the organization records no human judgment distinct from the machine's.*\n\n*Sequence matters.*\n\n*For consequential decisions, there are cases where the human view should be captured before machine advice is revealed. The point is not to isolate people from useful assistance or to romanticize unaided judgment. Humans bring bias, inconsistency, hierarchy, and self-protection into decisions. The point is to preserve a recoverable difference between human judgment and machine influence.*\n\n*Only then can the organization ask useful questions: Did the advice change the decision? Why? Was the change an improvement? Did the reviewer defer outside their competence? Did the same pattern recur across teams or agents?*\n\n*Meaningful human power is also more specific than universal approval. It belongs where people must define purpose, set boundaries, judge consequences, resolve conflicts of interest, authorize exceptions, and order correction or legitimate forgetting.*\n\n*The goal is not to keep a human hand on every task. It is to keep human authority at the points where society, law, and the organization expect someone to answer for the result.*\n\n**Human presence is not meaningful human power.**\n\n## From workflow automation to institutional architecture\n\n*The answer is not a larger approval matrix. Nor is it one monolithic platform through which every decision must pass.*\n\n*The human-agent enterprise needs three distinct planes.*\n\n*The****knowledge plane****governs claims, evidence, versions, standing, and correction. The****authority plane****governs roles, permissions, thresholds, exceptions, and obligations. The****assurance plane****tests whether the first two work as intended and whether the evidence is strong enough to justify continued reliance.*\n\n*Those planes must support three loops.*\n\n*The****production loop****creates and revises work. The****assurance loop****challenges claims, tests controls, and detects drift or failure. The****learning loop****ensures that verified corrections change future retrieval, evaluation, policy, and action.*\n\n*The learning loop is the one organizations most often neglect. A finding enters an audit report. A reviewer corrects an agent's output. A team discovers that a trusted source was wrong. The incident is closed — but the correction never reaches the prompts, retrieval paths, policies, or decision rules that will shape the next case.*\n\n*The enterprise observed an error. It did not learn from it.*\n\n*An assurance process that cannot change future behavior is not a control. It is documentation.*\n\n*This is why the AI-nterprise is an institutional architecture, not merely a technical stack. The controls may be implemented across identity systems, policy engines, knowledge graphs, audit records, workflow tools, and agent runtimes. What matters is not that one product owns the whole design. What matters is that the distinctions survive across the whole system.*\n\n## A manifesto for the Human-Agent Enterprise\n\n*The principles are simple to state. Their consequences are not.*\n\n*They are the ten principles of the white paper, stated here for leaders rather than for architects.*\n\n***1. Draft by default.****A generated claim is born without standing, without a place in memory, and without the right to trigger action. Nothing it does on its own changes that.*\n\n***2. Make promotion explicit.****A claim enters reusable memory only through an act someone owns, for a stated purpose, on a stated evidence basis, with a rule for when it leaves. “Accepted” is incomplete without “accepted for what.”*\n\n***3. Let evidence travel.****Qualifiers, uncertainty, counterevidence, transformations, and dependencies stay attached to the version. A paragraph copied into a new system does not become a new truth.*\n\n***4. Treat provenance as necessary but insufficient.****Knowing where a claim came from says nothing about whether it is warranted or who may act on it. A trace is not a warrant.*\n\n***5. Keep the four dimensions separate.****Visibility, standing, persistence, and action rest on distinct policy bases. Access does not confer truth, storage does not confer acceptance, and no system may manufacture its own authority: an agent acts only within a mandate granted elsewhere, and greater capability does not silently enlarge it.*\n\n***6. Scale scrutiny with consequence and persistence.****Expected harm, irreversibility, affected scope, uncertainty, and reuse set the intensity of control. Exploration and clinical action do not deserve the same gate.*\n\n***7. Preserve independent judgment when warranted.****Where consequences justify it, capture the human view before the advice, the change after it, and the reason for the change, so that judgment stays distinguishable from persuasion.*\n\n***8. Make challenge and removal first-class.****Contestation, supersession, retraction, repair, and forgetting are ordinary operations. A memory that cannot propagate corrections preserves error; one that cannot forget becomes liability.*\n\n***9. Require assurance to change the system.****Material findings enter owned remediation and later verification. Otherwise the same failure returns wearing a different sentence.*\n\n***10. Own the institutional learning loop.****Context, evaluations, corrections, policies, and governed memory must survive the substitution of any model. The model may be rented; the learning must be yours.*\n\n*These commitments are not anti-automation. They are what make deeper automation possible without making the institution unintelligible.*\n\n## What leaders should do now\n\n*The first step is not to buy another tool. It is to change the unit of analysis.*\n\n***Map decision surfaces, not only AI systems.****Identify where generated or machine-mediated claims acquire standing, enter memory, cross visibility boundaries, or trigger action. A low-profile summarizer may create more institutional risk than a famous model if its output quietly becomes precedent.*\n\n***Define claim and consequence classes.****Classify uses as ephemeral, bounded, material, or high-consequence — and prohibit some uses outright. Then specify the evidence, human authority, retention, and escalation each class requires.*\n\n***Separate the four decisions in policy and architecture.****Do not let storage stand in for acceptance, access stand in for authority, or a model score stand in for organizational warrant.*\n\n***Redesign human review around independent judgment.****Reserve decision-before-advice, recorded rationale, and contestability for the points where anchoring or deference could materially change an outcome. Do not waste scarce human attention on ritual clicks.*\n\n***Close the assurance-learning loop.****Test whether a verified correction changes what the next person or agent will retrieve, treat as warranted, and be allowed to do. If it does not, the organization is accumulating reports rather than knowledge.*\n\n*None of this requires waiting for autonomous agents to become perfect. It requires accepting that imperfect agents are already entering imperfect institutions — and that the combination will not govern itself.*\n\n## The enterprise after software\n\n*The internet made the enterprise networked. Cloud made infrastructure elastic and software increasingly rented. AI is doing something more radical: it is distributing the production of judgment.*\n\n*That creates extraordinary leverage. It also removes many of the frictions that once made authority visible. The memo had an author. The policy had an owner. The analyst could be questioned. The system executed a rule someone had written. In an agentic workflow, evidence selection, interpretation, recommendation, execution, and memory can happen inside a chain that no participant sees in full.*\n\n*The answer is not to freeze the enterprise in the human-only past. It is to design an institution capable of carrying distributed intelligence without losing the ability to explain itself.*\n\n*The defining question is no longer how much AI the enterprise uses. It is whether the enterprise can still tell how something became known, who was allowed to act on it, what changed when it was wrong, and who remained accountable throughout.*\n\n*An enterprise with agents does not yet have an AI-nterprise.*\n\n*It has one when intelligence can be distributed without making authority, accountability, and memory disappear.*\n\n*This article introduces the argument developed in the open white paper****AI-nterprise: Governing What the Organization Knows****(2026):**[https://doi.org/10.5281/zenodo.22282125](https://doi.org/10.5281/zenodo.22282125)**.**The white paper credits the research this argument builds on and states what is and is not new here.*\n\n*I am continuing to explore these themes through research and advisory work. If your leadership team is redesigning how humans and agents share knowledge, judgment, and authority, I am open to those conversations.*\n\n© 2026 Ottaviano De Cicco.\n\n[Use the Four Questions worksheet](https://ai-nterprise.com/tools/#four-questions) [Explore the book](https://ai-nterprise.com/book/)\n" }, { "path": "/articles/the-human-hallucination/", "title": "The Human Hallucination: The AI Failure Mode Nobody Is Measuring", "description": "Why human judgment and confidence must be assessed alongside machine output.", "content_status": "published", "h1": "The Human Hallucination: The AI Failure Mode Nobody Is Measuring", "canonical": "https://ai-nterprise.com/articles/the-human-hallucination/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/articles/the-human-hallucination.md", "html_sha256": "cf9477e8fef99be82d4447481aa2f4c4eca0416555f3eb877313ededba53a3e5", "markdown_sha256": "c2233989297c48a91b8865dc7e8735c71717713665159260f2a34369591b3104", "content_markdown": "Canonical: https://ai-nterprise.com/articles/the-human-hallucination/\nLanguage: en\nContent status: published\n\n[Ideas](https://ai-nterprise.com/articles/) · Judgment\n\n# The Human Hallucination: The AI Failure Mode Nobody Is Measuring\n\nOttaviano De Cicco · 12 June 2026 · 10 min read\n\nFirst published on [LinkedIn ↗](https://www.linkedin.com/pulse/human-hallucination-ai-failure-mode-nobody-measuring-de-cicco-exjhf), 12 June 2026.\n\n![](https://ai-nterprise.com/assets/editorial-judgment.webp)\n\n*We built leaderboards for machine hallucination. The more dangerous miscalibration is happening on our side of the screen.*\n\nA friend of mine is a hematologist. Thirty years of clinical practice, two degrees, a career measured in the patients he has pulled back from the edge. Last week he told me something that has stayed with me.\n\n\"Something has changed in my consultations,\" he said. \"I deliver a diagnosis, and increasingly the patient pushes back: *Doctor, couldn't it also be this other condition? Isn't there a different medication for my case?* They have already asked ChatGPT. They arrive with the answer.\"\n\nHe was careful to add — and this matters — that he is not complaining about informed patients. Informed patients ask better questions, catch real errors, and take better care of themselves. Medicine's paternalistic past was not a golden age. What troubles him is something more specific: the *weight* his patients now assign to the chatbot's answer. Not \"here is something I read, what do you think?\" but a quiet conviction that the machine's output and his clinical judgment sit at the same level — two opinions, equally weighted, one of which happens to be free and available at 3 a.m.\n\nThis is not a fringe behavior anymore. A KFF tracking poll conducted in early 2026 found that 32% of American adults had turned to AI chatbots for health information in the past year — roughly the same share that relies on social media for health advice. Among those users, 41% had uploaded personal medical records — test results, doctors' notes — to get personalized explanations. A Rock Health survey from December 2025 found that consumer use of AI for health questions had doubled year over year, with the most common queries being treatment options for a known diagnosis and symptom research. The patient with the chatbot is not an anecdote. She is the new baseline.\n\n## Naming the second hallucination\n\nThe AI industry spends enormous and growing energy measuring **machine hallucination** — the tendency of models to state falsehoods with perfect fluency. Public leaderboards rank models by hallucination rate. Evaluation suites probe factuality and calibration. This work is necessary, and it is accelerating.\n\nBut there is a second hallucination, and almost nobody is measuring it.\n\nCall it **human-AI hallucination**: the systematic inflation of a person's confidence in their own knowledge, induced by interaction with an AI. The machine hallucinates facts. The human hallucinates competence. And the second failure mode is arguably more dangerous than the first, because it has no leaderboard, no benchmark, and no warning label — and because it compounds silently inside decisions that look, from the outside, perfectly well-informed.\n\nThree mechanisms produce it. None of them is the user's fault.\n\n## Mechanism one: the flattened voice\n\nFor the entire history of our species, fluency has been a serviceable proxy for competence. The person who can explain a subject clearly, confidently, and in well-formed sentences usually understands it. We are wired to read epistemic confidence from linguistic confidence, and for millennia that heuristic worked.\n\nA language model breaks it. The model delivers settled scientific consensus and plausible confabulation **in the same voice** — identical syntax, identical confidence, identical typography. Thirty years of hematology literature and a statistically shaky association arrive on the patient's screen wearing the same suit. The interface flattens epistemic status: there is no perceptible difference between \"this is textbook medicine\" and \"this is a pattern I assembled that sounds right.\"\n\nSeen this way, the patient in my friend's office is not overestimating herself. She is importing the machine's *apparent* confidence without any view into its *actual* calibration — because the product gives her none. The hallucination is not a character flaw. It is an interface outcome.\n\n## Mechanism two: the validation asymmetry\n\nA diagnosis resists you. The doctor does not particularly care whether you like what he found; the lab values are what they are. That resistance — the friction of an answer that does not bend to your preferences — is a feature of expertise, not a bug of bedside manner.\n\nA chatbot, by design and by training, is agreeable. It tends to adopt the frame of whoever is asking. Ask it whether your symptoms could be something less frightening, and it will explore that possibility with you, thoroughly and kindly. The patient does not merely receive information; she receives **validation**, shaped to the contours of the question she chose to ask.\n\nThis is also why the phenomenon is materially worse than the \"Dr. Google\" era we already lived through. Search returned ten contradictory links and left the synthesis — the hard part — to you. The friction was annoying, and it was also a safety mechanism: disagreement between sources signaled uncertainty. The chatbot returns one answer — coherent, fluent, tailored, and pleasant. The synthesis has been done for you, invisibly, by a system optimized to be helpful to *you specifically*. Friction was a feature. We removed it.\n\n## Mechanism three: the jagged frontier\n\nIn 2023, a team of researchers from Harvard, Wharton, MIT, and BCG ran one of the most cited field experiments on AI and knowledge work (Dell'Acqua et al., \"Navigating the Jagged Technological Frontier\"). Hundreds of consultants worked with and without AI assistance. The headline finding was nuanced in a way the headlines were not.\n\nOn tasks **inside** the model's frontier of competence, AI helped everyone — and it helped the weakest performers most. Execution gaps compressed. This is the genuinely good news, and it is real.\n\nBut on tasks **beyond** the frontier — tasks that looked similar but required judgment the model did not have — consultants using AI performed markedly *worse* than those without it: roughly nineteen percentage points less likely to reach the correct answer. They trusted polished output on a task the machine could not actually do, and the polish concealed the failure.\n\nHere is the asymmetry that matters: detecting a wrong answer requires precisely the expertise you are delegating. AI compresses execution gaps and **amplifies judgment gaps**. The person with deep domain knowledge uses the model as an instrument and catches its failures. The person without it uses the model as an oracle and inherits its failures — at scale, with confidence. A sharp mind with AI gets sharper. A fool with AI becomes a fool at scale.\n\n## What experience actually is\n\nThis is the distinction my friend the hematologist was reaching for, and it deserves to be said plainly.\n\n**Experience is knowledge that has paid a price.** His thirty years are not worth more than the model's training data because they contain more facts — they do not; the model holds more facts than any human who has ever lived. They are worth more because every pattern he carries has been tested against consequences: real patients, real errors, real complications, real nights spent wondering whether the call he made at 6 p.m. was the right one. His knowledge has skin in the game. When he is wrong, he signs the chart, attends the follow-up, carries the outcome.\n\nThe model's knowledge has never paid anything. If it is wrong about the patient in front of him, it does not revise itself, does not lose sleep, does not testify. It will give the next patient the same confident answer.\n\nAnd there is a clinical corollary that no benchmark captures: the model answers the question **as posed**. The expert questions the question. The professor looks at the patient — her history, her gait, the thing she mentions on the way out the door — and sees what was never in the prompt. Half of diagnostic skill is knowing which question the situation is actually asking. That skill does not transfer through a chat window, because it operates *before* the chat begins.\n\n## We have seen this movie — but the sequel is darker\n\nIf you were working in the late nineties, you remember the overnight epidemic of \"web experts.\" A technology arrived, careers were at stake, and résumés adapted faster than skills did. Browse LinkedIn today and you will see the rerun: AI skills everywhere, on profiles that a year ago contained none. Self-marketing is human, and mostly harmless.\n\nBut there is a disanalogy, and it is the uncomfortable core of this piece. In 1999, people overclaimed mastery of a **tool**. Today, people overclaim possession of **knowledge itself** — because the tool serves it up on demand, fluently, to anyone. It is the difference between owning a library and having read it. Access to knowledge has never been cheaper; the ability to evaluate it has never been more unevenly distributed. The gap between the two is exactly where human-AI hallucination lives.\n\n## What this means in practice\n\nNot \"trust the experts and stop asking questions.\" That conclusion would be wrong on the merits — patient pushback catches real errors, and expert authority has its own failure modes — and futile besides: the chatbot-assisted patient, employee, and customer are permanent features of the landscape now.\n\nThe practical conclusions are narrower and more useful:\n\n**Treat AI output as an unsigned draft.** Fast, useful, and unaccountable until a human with verification capacity signs it. The signature is not a formality; it is the moment accountability enters the process. Any workflow where AI output flows into consequential decisions without a qualified signature is accumulating silent risk.\n\n**Hire and promote for verification capacity, not tool usage.** \"Uses AI\" is no longer a differentiator; within a year it will be as informative as \"uses email.\" The differentiating skill is the ability to know when the output is wrong — which is, inconveniently, the old skill: domain expertise, now repriced upward precisely when the market believes it has been made obsolete.\n\n**Keep judgment where accountability lives.** The right division of labor is emerging clearly from the research: let AI raise the floor of execution, and keep human judgment at every point where someone must answer for the outcome. Organizations that invert this — automating judgment while keeping humans on execution — are building the failure mode into the org chart.\n\n**If you are the expert: coach calibration, not compliance.** The answer to the patient who arrives with a chatbot transcript is not dismissal. It is the sentence my friend has started using, which I think is exactly right: *\"Good — now let me show you what the model cannot see about you.\"*\n\n## The measurement gap\n\nIn a previous article I argued that the missing layer of the AI economy is independent trust infrastructure — credible, third-party ways to verify what an AI system actually knows and where it fails. That work is beginning: machine hallucination is now measured, ranked, and published.\n\nBut the dyad — the human plus the machine, the unit that actually makes decisions in the real world — has no calibration score. Nobody measures how much an AI inflates its user's confidence beyond their competence, even though that number, if we had it, would explain more about AI-era decision quality than any model benchmark. The machine's hallucination and the human's are the same failure observed from two sides of the screen. Until we measure both, every claim about AI productivity is a numerator without a denominator.\n\nThe hematologist will keep practicing medicine the only way it can be practiced: one accountable judgment at a time. The open question — for every leadership team deploying these tools — is whether your organization still knows which of its judgments are accountable, and which have quietly become unsigned drafts.\n\n*I am continuing to explore these themes through advisory work and ongoing research. If your leadership team is thinking seriously about where human judgment must sit in an AI-augmented organization — and how to verify what your AI systems actually know — I am open to those conversations.*\n\n**Sources:** KFF Tracking Poll on Health Information and Trust, fielded February 24 – March 2, 2026 (published March 25, 2026); Rock Health Digital Health Consumer Adoption Survey, December 2025; F. Dell'Acqua et al., \"Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality,\" Harvard Business School Working Paper 24-013 (2023), published in *Organization Science* (2026).\n\n© 2026 Ottaviano De Cicco.\n\n[Use the Four Questions worksheet](https://ai-nterprise.com/tools/#four-questions) [Explore the book](https://ai-nterprise.com/book/)\n" }, { "path": "/articles/the-model-is-rented/", "title": "The Model Is Rented. The Learning Must Be Yours.", "description": "How an organization retains what it learns while using rented models.", "content_status": "published", "h1": "The Model Is Rented. The Learning Must Be Yours.", "canonical": "https://ai-nterprise.com/articles/the-model-is-rented/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/articles/the-model-is-rented.md", "html_sha256": "f24b6376a49ba9a279e3430d2d55201619eb12a4d15d8cfa0db42dc27b999b84", "markdown_sha256": "61ec2fa0ed1bb9951667e3b4431fec418a71de45c7019b6333b629073b076c18", "content_markdown": "Canonical: https://ai-nterprise.com/articles/the-model-is-rented/\nLanguage: en\nContent status: published\n\n[Ideas](https://ai-nterprise.com/articles/) · Learning\n\n# The Model Is Rented. The Learning Must Be Yours.\n\nOttaviano De Cicco · 15 July 2026 · 11 min read\n\nFirst published on [LinkedIn ↗](https://www.linkedin.com/pulse/model-rented-learning-must-yours-ottaviano-de-cicco-16xsf), 15 July 2026.\n\n![](https://ai-nterprise.com/assets/editorial-learning.webp)\n\n*Satya Nadella's \"Reverse Information Paradox\" identifies the enterprise AI asset most companies are failing to protect.*\n\nMost companies have an AI budget. Almost none has a learning balance sheet.\n\nOn July 12, 2026, Microsoft CEO Satya Nadella published a long-form post on X titled *The Reverse Information Paradox*. He starts with Kenneth Arrow's classic information paradox: a seller cannot demonstrate the value of information without revealing it, but once revealed, the buyer already possesses it.\n\nAI reverses the direction of exposure.\n\nTo make an external model useful, the buyer must provide context: documents, workflows, exceptions, corrections, evaluation criteria, and the institutional knowledge that distinguishes one company from another. The more useful the system becomes, Nadella argues, the more the organization reveals about how it thinks. His conclusion is memorable: \"In consuming intelligence, you are creating intelligence. And what you create should belong to you.\"\n\nIt is a sharp observation. It is also easy to misunderstand.\n\nThe quickest interpretation is a familiar one: do not upload confidential documents, negotiate no-training clauses, deploy open models on-premise, and keep sensitive data behind the firewall. All sensible measures. All incomplete.\n\nThe strategic question is not only where your data travels. It is where your learning accumulates.\n\n## The tension inside Nadella's paradox\n\nThe paradox describes a real asymmetry: the provider can progressively learn how your business operates, while you have little visibility into what it learns in return. Read as a claim about foundation-model training, however, it requires an important qualification.\n\nMicrosoft's own documentation states that prompts, completions, embeddings, and training data submitted to Models sold by Azure in Microsoft Foundry — including Azure OpenAI models — are not used to train foundation models without the customer's permission or instruction. OpenAI makes a similar commitment for its business products and API platform: business data is not used for model training by default. These protections do not eliminate every privacy, retention, security, or contractual risk — those remain dependent on the product, configuration, and agreement being used. But they do change the diagnosis: it is not accurate to assume that every enterprise prompt automatically enters the next training run of a frontier model.\n\nSo how can the enterprise still pay twice? Because foundation-model training is not the only place where organizational learning can be lost.\n\nAn employee asks an AI to prepare a proposal. The first version is wrong. A senior colleague corrects the positioning, removes an argument that does not work in that industry, changes the pricing logic, and explains which objection the client is really likely to raise. The final proposal is excellent. The task has been completed.\n\nBut where did the correction go?\n\nPerhaps it remains inside one conversation. Perhaps it disappears into an employee's personal prompt history. Perhaps the next salesperson starts again from zero and makes the same mistakes.\n\nThe vendor may not have learned anything from that interaction.\n\nBut neither has the company.\n\nCall this learning leakage: the value created during human-AI interaction is not converted into a reusable, portable, organization-owned asset. Sometimes knowledge leaks outward. More often, it simply evaporates.\n\nAn organization does not need to have its knowledge stolen in order to lose it. It only needs to fail to retain it.\n\n## Every AI interaction produces two outputs\n\nThe first output is visible. It is the report, the code, the forecast, the customer response, the legal summary, or the product recommendation the user requested. That output completes today's task.\n\nThe second output is less visible. It is the record of what the user asked, which context mattered, which tools were selected, what the model misunderstood, what the expert corrected, which answer was rejected, how success was evaluated, and where human escalation became necessary. That output determines whether tomorrow's task will be performed better.\n\nThe first is usually counted as productivity. The second is rarely accounted for at all.\n\nMost enterprise AI programs are designed to generate outputs. Very few are designed to capture the learning created while those outputs are being produced. This is the asset Nadella's essay points toward, even though the distinction deserves to be made more explicit: the real second payment is not always knowledge transferred to the provider. It is often knowledge created by the enterprise but never added to the enterprise.\n\n## Data is not the same as learning\n\nCompanies have spent two decades building data estates: customer records, product catalogs, contracts, transaction histories, policies, telemetry, and documents. That data is essential. But data describes what the organization knows. Learning describes how the organization decides.\n\nA policy document tells the model what the official rule is. An expert correction reveals when the rule should not be applied literally. A CRM database shows which deals closed. A senior salesperson's feedback explains why one argument worked and another damaged credibility. A risk model provides a probability. An experienced underwriter's decision reveals when the available variables fail to capture the real exposure. A product specification describes what can be built. An engineering escalation reveals which compromise the company is willing to accept when time, cost, safety, and customer expectations collide.\n\nThe documents contain information. The exceptions contain judgment.\n\nPrompts reveal what people are trying to accomplish. Corrections reveal what the organization considers wrong. Private evaluations reveal what the organization considers good. Escalations reveal where accountability still requires human expertise. Together, these form something far more valuable than a prompt library: they form the organization's operational intelligence.\n\nA competitor can license the same foundation model. It may even acquire similar public data. What it cannot easily buy is twenty years of accumulated exceptions, trade-offs, rejected alternatives, and accountable decisions.\n\nThe competitive asset is not the answer. It is the system that becomes better at recognizing what counts as a good answer.\n\n## The Enterprise Learning Loop\n\nA functioning enterprise learning loop has a simple sequence:\n\nContext. Execution. Evaluation. Correction. Adaptation. Memory. Reuse.\n\nThe model receives the relevant context and performs a task. A person or automated control evaluates the result. Errors and exceptions are captured. The workflow, retrieval layer, instructions, tools, or adapted model are improved. The approved learning enters organizational memory. The next execution begins from a higher level.\n\nThe loop compounds only when the correction changes future performance. If an expert corrects an answer and the improvement affects only that conversation, the company has purchased a useful service. If the correction improves every relevant future interaction, the company has built a capability.\n\nThat capability depends on several assets. Private evals encode what good performance means inside the organization. Traces preserve the context, tool use, sequence, and provenance behind an outcome. Structured feedback captures not only that an answer was wrong, but why. Organizational memory retains approved knowledge without turning every historical interaction into uncontrolled context. Orchestration determines which model, data source, tool, and human authority should handle each part of the workflow. Adaptation rights allow the organization to use its outputs, corrections, and evaluations to improve its own systems.\n\nThese are not secondary pieces of telemetry. They are operating intellectual property. The enterprise does not necessarily need to own the foundation model, the GPUs, or every component of the infrastructure. It needs to own the layer that remembers why one answer was accepted and another was rejected.\n\nOwn the loop, not necessarily the model.\n\n## The model-removal test\n\nThere is a practical way to determine whether an organization owns its learning. Imagine that its primary AI model or provider becomes unavailable on Monday morning.\n\nCan the company still run its private evaluations? Can it export and reuse its organizational memory? Can its workflows be directed toward another model without being rebuilt from scratch? Do its corrections, decision rules, prompt patterns, and escalation logic remain usable? Can it reproduce why important outputs were approved? Will the replacement system begin as a novice, or inherit the experience accumulated by its predecessor?\n\nIf removing one model also removes the organization's acquired competence, the company has not created an enterprise AI capability. It has created a dependency.\n\nLegal ownership of inputs and outputs is important, but it is not sufficient. Operational ownership means being able to inspect, export, reuse, adapt, and continue learning when a supplier changes.\n\nNadella recommends decoupling orchestration from any single model. That is exactly the right principle — and it should apply to Microsoft just as rigorously as it applies to OpenAI, Google, Anthropic, or any open-model provider.\n\nA principle of model independence is valuable only when it includes independence from the company proposing it.\n\n## Cloud versus on-premise is the wrong argument\n\nThe immediate response to this problem is often to frame it as a deployment choice: cloud or on-premise, closed or open source, API or owned infrastructure. These choices matter, but they answer different questions.\n\nOn-premise deployment answers where computation occurs. An open-weight model makes its parameters available for local operation or adaptation; its license determines what an organization may legally do with them. Neither fact, by itself, guarantees that the complete system qualifies as open source — a distinction formalized by the Open Source Initiative. And neither decision, by itself, answers who owns the evaluations, memory, corrections, workflow logic, and accumulated judgment.\n\nA cloud architecture can preserve learning sovereignty when it provides strong tenant boundaries, explicit no-training commitments, usable export paths, customer-controlled memory, and model-independent orchestration. An on-premise architecture can still destroy it when evaluations live in scattered spreadsheets, corrections remain inside individual chat sessions, logs are inaccessible, and the entire workflow depends on one proprietary stack.\n\nThe same is true of cost. Open models and self-hosted infrastructure can produce major savings in the right workload. But there is no universal 90% rule. One recent preprint estimates break-even periods ranging from months to years depending on model size and volume, while another finds that low utilization alone can increase the effective cost of self-hosted inference by more than twentyfold. \"90% cheaper\" may be the result of a particular benchmark. It is not an enterprise architecture.\n\nDeployment location is an implementation choice. Learning sovereignty is an operating-model choice.\n\n## What enterprise leaders should do now\n\nFive actions should be on the leadership agenda for the next two quarters.\n\n**Inventory the second output.** Extend the traditional data inventory to include prompt patterns, evaluations, corrections, decision traces, agent tool calls, memories, fine-tuned components, workflow rules, and escalation logic. Identify where each asset resides, who can access it, how long it is retained, and whether it can be exported in a usable form.\n\n**Treat private evals as core intellectual property.** Public benchmarks tell you whether a model is broadly capable. They do not tell you whether it performs correctly inside your organization. The scenarios, thresholds, edge cases, and failure conditions used to evaluate AI should be versioned, governed, and owned like any other strategic asset. Whoever defines the evals defines what good means.\n\n**Capture expert correction at the point of work.** Do not expect senior professionals to maintain a separate knowledge-management system after completing the task. Build feedback capture into the workflow: what was accepted, what was rejected, what changed, why it changed, and what outcome followed. Expert attention is expensive. Paying for the same correction repeatedly is one of the least visible costs in enterprise AI.\n\n**Separate the learning layer from the model layer.** Evals, memory, workflow state, permissions, provenance, and business rules should not exist only inside a model provider's interface. Architect them as enterprise-controlled services that can work across multiple models. Then test that portability before a commercial or operational crisis makes the test unavoidable.\n\n**Negotiate operational ownership — and measure compounding.** Contracts should address more than input and output ownership. They should clarify retention, feedback use, deletion, portability, adaptation and fine-tuning rights, subprocessors, model substitution, and exit assistance. Internally, measure whether corrected errors recur, how quickly verified learning is reused, how performance changes against private evals, and what it would cost to switch providers.\n\nAI productivity is not compounding merely because usage is increasing. It compounds when every verified correction makes the organization permanently better.\n\n## The moat is moving\n\nAccess to a powerful model can provide an advantage today. But access diffuses. Competitors can subscribe to the same APIs, deploy the same open-weight models, hire from the same labor market, and connect similar enterprise data. The durable difference will be the speed at which each organization converts experience into retained intelligence.\n\nTwo companies can use the same model on similar documents. One company's experts repeatedly correct the same errors in isolated conversations. The other captures those corrections, converts them into evaluations and workflow improvements, tests them, and distributes the resulting learning across the organization.\n\nAfter one interaction, the difference is small. After one million interactions, they are no longer using the same system in any meaningful sense.\n\nOne has an AI tool. The other has an institutional learning machine.\n\n## Trust, judgment, ownership\n\nIn *The Trust Gap*, I argued that the emerging agent economy requires independent ways to verify what AI systems actually do. In *The Human Hallucination*, I argued that the human-machine pair must also be evaluated, because AI can increase a person's confidence faster than it increases their competence. The Reverse Information Paradox introduces the third layer.\n\nTrust asks whether the system performs as claimed.\n\nJudgment asks whether the human knows when it does not.\n\nOwnership asks whether each verified correction makes the organization permanently smarter.\n\nThese are not separate governance problems. They form the control system of the AI-augmented enterprise.\n\nThe board-level question should therefore no longer be only: *Which model are we buying?* It should be: *When our people teach an AI how this company works, who gets smarter permanently?*\n\nRent the model when that is efficient.\n\nNever rent your organization's learning.\n\n**Sources:** Satya Nadella, \"The Reverse Information Paradox,\" long-form post published on X, July 12, 2026; Kenneth J. Arrow, \"Economic Welfare and the Allocation of Resources for Invention,\" in *The Rate and Direction of Inventive Activity* (1962); Microsoft, \"Data, privacy, and security for Foundry Models sold by Azure\"; OpenAI, \"Enterprise Privacy at OpenAI\"; Open Source Initiative, \"The Open Source AI Definition 1.0\"; G. Pan, V. Chodnekar, A. Roy, and H. Wang, \"A Cost-Benefit Analysis of On-Premise Large Language Model Deployment: Breaking Even with Commercial LLM Services,\" arXiv:2509.18101 (2025, preprint); C. Patil, \"Beyond Per-Token Pricing: A Concurrency-Aware Methodology for LLM Infrastructure Cost Estimation,\" arXiv:2606.11690 (2026, preprint).\n\n*Author's note: I am continuing to explore these themes through advisory work and ongoing research. I welcome conversations with leadership teams examining how to preserve institutional judgment, model choice, and learning sovereignty as AI moves from experimentation into core operations.*\n\n© 2026 Ottaviano De Cicco.\n\n[Use the Four Questions worksheet](https://ai-nterprise.com/tools/#four-questions) [Explore the book](https://ai-nterprise.com/book/)\n" }, { "path": "/book/", "title": "The book", "description": "AI-nterprise by Ottaviano De Cicco: the argument, open white paper and free printable resources.", "content_status": "published", "h1": "AI-nterprise", "canonical": "https://ai-nterprise.com/book/", "share_image": "https://ai-nterprise.com/assets/share-book.png", "markdown_url": "https://ai-nterprise.com/agents/content/book.md", "html_sha256": "c6b5eaecc1e0f60215577b16a1b83dcbc4ae1707d71800396fb6f502b8489f6c", "markdown_sha256": "63713fa20036c73fe57ea8aa38e0fed4b606fce3ff3141d17eceb727b79352bb", "content_markdown": "Canonical: https://ai-nterprise.com/book/\nLanguage: en\nContent status: published\n\nThe book / Ottaviano De Cicco\n\n# AI-nterprise\n\nGoverning Knowledge, Judgment, and Authority in the Age of AI Agents\n\nI ask how an enterprise can govern what it knows, remembers, and allows to act.\n\nForthcoming · Q4 2026 · English edition\n\n[Get the free worksheet](https://ai-nterprise.com/book/#resources) [Read the white paper ↗](https://doi.org/10.5281/zenodo.22282125)\n\n![AI-nterprise cover: pale background, deep green AI and central node, dark title and pathways](https://ai-nterprise.com/assets/book-cover.png)\n\nWho this book is for\n\n## The person who signs.\n\nThis book is for the person who signs for what the organization does with what it believes: a chief executive, a chief operating, risk or data officer, the head of a function where agents already draft, recommend or act. It assumes no technical background. It assumes responsibility.\n\nWhat you take away\n\n## A method for one workflow.\n\nA vocabulary for decisions that today hide inside the word “approved”. Ten principles. A method you can start on one workflow next Monday, with the worksheet available on this page. And a way to tell, after the model changes, what your organization has actually learned.\n\nThe argument\n\n## Follow the answer after it leaves *the conversation.*\n\nAn agent produces an answer. Someone copies it into a report. Another system retrieves it. A team acts on it. AI-nterprise follows the organizational claim through those handoffs and asks where its authority comes from.\n\nThe book examines how generated output becomes warranted knowledge, how memory shapes later work, and how a recommendation acquires permission to change something outside the conversation.\n\nIts central question is practical: can the enterprise identify what it relied on, who authorized the consequence, and what must change when the underlying claim is corrected?\n\nInside the book\n\n## Follow the argument. *Into the enterprise.*\n\nFifteen chapters, from the arrival of a new organizational actor to the work of keeping the architecture accountable.\n\n### Explore the fifteen chapters\n\n- 01 From e-Enterprise to AI-nterprise\n\n- 02 A New Actor in the Organization\n\n- 03 The Unsigned Draft\n\n- 04 Trust Without Self-Certification\n\n- 05 The Human Hallucination\n\n- 06 The Powers Humans Must Keep\n\n- 07 The Learning Must Be Yours\n\n- 08 Memory Is a Decision\n\n- 09 Four Decisions Hiding Inside \"Approved\"\n\n- 10 Planes, Loops, and Gates\n\n- 11 Ten Rules for the AI-nterprise\n\n- 12 Start Where Consequence Lands\n\n- 13 Can the Enterprise Correct Itself?\n\n- 14 Keeping the Architecture Working\n\n- 15 The Enterprise After Software\n\nThe open white paper\n\n## The reference model *and its limits.*\n\nRead the model behind the book in the open white paper.\n\nVersion 1.0 was published on 3 September 2026 and is openly available on Zenodo.\n\n[Read the white paper ↗](https://doi.org/10.5281/zenodo.22282125)\n\nResources for the reader\n\n## One worksheet. *Four distinct decisions.*\n\nOne open download. No account or signup.\n\n02 / Four organizational decisions\n\n### Four Questions Before an Agent's Answer Becomes Action\n\nUse this worksheet on one or two real processes in which an agent produces, transforms, recommends, or passes on a claim.\n\nPDF · 2 pages · v0.1 preview · CC BY-SA 4.0\n\nInstructions, process map and filled example.\n\n[Download the worksheet](https://ai-nterprise.com/book/downloads/four-questions.pdf)\n\nWorksheet content © 2026 Ottaviano De Cicco. [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/). The AI-nterprise name and trademark are excluded from this licence; attribution to the author is required.\n" }, { "path": "/contact/", "title": "Contact Ottaviano De Cicco", "description": "Contact Ottaviano De Cicco about AI-nterprise and the governance of enterprise AI workflows.", "content_status": "published", "h1": "Let’s start a conversation.", "canonical": "https://ai-nterprise.com/contact/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/contact.md", "html_sha256": "e891ae4c7d5573f60240ddc59c26ea94df393e707e7dceb91fb0c3489f051887", "markdown_sha256": "6d581d15ba1028bf1ebff671cfb680dba115aa70afdf303d7674fde2fbd6fc72", "content_markdown": "Canonical: https://ai-nterprise.com/contact/\nLanguage: en\nContent status: published\n\nContact · Ottaviano De Cicco\n\n# Let’s start a conversation.\n\nGet in touch about AI-nterprise or a decision your organization is facing with AI.\n\n## Contact me directly.\n\n[Contact Ottaviano on LinkedIn ↗](https://www.linkedin.com/in/ottavianodecicco)\n\nIf you are bringing a workflow, tell me which process you own, what the agent contributes, which decision follows, and where responsibility or correction becomes unclear.\n\n## A useful starting point.\n\nThe Four Questions worksheet helps you describe one process before the conversation. Download it freely, without an account or signup.\n\n[Use the Four Questions worksheet](https://ai-nterprise.com/tools/#four-questions)\n" }, { "path": "/framework/", "title": "Framework · A visual reference model", "description": "Follow the AI-nterprise model visually: four decisions, three control planes and three coupled loops.", "content_status": "published", "h1": "A claim becomes a consequence.", "canonical": "https://ai-nterprise.com/framework/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/framework.md", "html_sha256": "8f070c8fa1719ba911023dd9ceb56c2d850dee9b035b4eee09f495a09cf1bf59", "markdown_sha256": "c14ee6dd16a64e09824fb07949b7eae99df294738e0bb4b563c951b4c41496c2", "content_markdown": "Canonical: https://ai-nterprise.com/framework/\nLanguage: en\nContent status: published\n\nFramework · The public foundation\n\n# A claim becomes *a consequence.*\n\nFollow one claim through four decisions, three control planes and three continuing loops.\n\nOn this page\n\n[The moment of consequence](https://ai-nterprise.com/framework/#foundation) [Four decisions](https://ai-nterprise.com/framework/#four-decisions) [Three planes](https://ai-nterprise.com/framework/#three-planes) [Three loops](https://ai-nterprise.com/framework/#three-loops) [One case](https://ai-nterprise.com/framework/#one-case) [Ten principles](https://ai-nterprise.com/framework/#ten-principles) [Use and attribution](https://ai-nterprise.com/framework/#attribution)\n\n00 / The moment of consequence\n\n## Between an answer and an action.\n\n**Diagram transcript:** The moment of consequence · A specific claim version reaches a gate. Knowledge and Authority converge on the same gate. One conditional route leads to an effect; another holds the proposed use for review. Assurance probes Knowledge, Authority and the gate and sends findings to the correction owner. No assurance connection enters the action route. · What is claimed · What is decided · What happens · v3 · Claim · a specific version · Four independent decisions: standing, persistence, visibility, action enablement · Gate · If enabled for this use · C2 · Effect within scope · Knowledge · Evidence · standing · Authority · Mandate · limits · Use held · Proposal for review · Assurance · independent checks · Owner · correction\n\n**The moment of consequence.** Knowledge and authority meet at use time. A held use leaves the stored claim version intact. Four independent decisions: standing · persistence · visibility · action enablement.\n\n### AI-nterprise: Governing What the Organization Knows\n\nWhite paper · Version 1.0 · 3 September 2026\n\n[White paper · v1.0 ↗](https://doi.org/10.5281/zenodo.22282125) [Read the manifesto](https://ai-nterprise.com/articles/manifesto/)\n\n01 / Four dimensions\n\n## One claim, four decisions.\n\nStanding, retention, access and action each require their own decision.\n\nv3\n\nOne claim version\n**Four separate decisions**\n\n01 / Standing\n\n### Warranted for this use?\n\nPurpose and consequence determine the required support.\n\n02 / Persistence\n\n### Kept ≠ eligible\n\nIllustrative snapshot · one defined use\n\n| Version | Kept | Eligible |\n| --- | --- | --- |\n| v1 | Retained | No |\n| v2 | Retained | No |\n| v3 | Retained | Yes* |\n\n*Within its current scope and validity. History has its own retention rule.\n\n03 / Visibility\n\n### Who may retrieve which form?\n\nAccess includes the representation and its supporting context.\n\n04 / Action enablement\n\n### Who may act, within which boundary?\n\nA valid mandate and current conditions govern the effect.\n\nSeeing is not believing. Remembering is not accepting. Accepting is not authorizing action.\n\n[Use the Four Questions worksheet](https://ai-nterprise.com/tools/#four-questions)\n\n02 / Three control planes\n\n## Three planes. One moment of use.\n\n**Diagram transcript:** Three planes, one moment · Knowledge and Authority retain separate write authority and meet at use time. Crossed arrows forbid evidence from writing authority and a grant from manufacturing standing. Independent Assurance checks both planes and the gate, routing findings to the correction owner, never to the action path. · Knowledge · Claims · evidence · versions · standing · correction · Authority · Mandates · budgets · permissions · revocation · Four independent decisions: standing, persistence, visibility, action enablement · Gate · Use time · Action · Separate write authority · Neither rewrites the other · Assurance · tests both planes and the gate · Owner · correction\n\nNo single platform needs to own all three planes.\n\nEvidence may not write authority.\n\nA grant may not manufacture standing.\n\nPolicy may make action conditional on the claim’s current standing. Each plane retains its own write authority.\n\n### A failed condition and a revoked mandate\n\nA failed condition requires a fresh evaluation. A revoked grant requires a new act by the authority entitled to issue it.\n\n03 / Three coupled loops\n\n## Work. Challenge. Learn.\n\n**Diagram transcript:** Three loops, coupled · A production loop creates work and decisions. An assurance loop challenges and tests the claim and gate. A verified correction passes through explicit authorized promotion before learning changes the next relevant case. Loops are coupled processes, distinct from the control planes. · Production · Work · recommendations · decisions · revisions · v3 · Four independent decisions: standing, persistence, visibility, action enablement · Gate · C2 · Next · case · Assurance · Challenge · evaluate · detect · Verified correction · Authorized owner · promotes the change · Learning · Changed retrieval · evaluation · policy · memory · action\n\nLoops describe continuing processes. Planes describe control responsibilities; there is no one-to-one mapping.\n\nIf a verified correction does not alter the next relevant case, the enterprise observed an error but did not learn.\n\n04 / An illustrative case · Chapter 9\n\n## Kept in history. Held from a new use.\n\nNew customer accounts require review of an earlier liquidity assessment. Its earlier version remains available to explain the past decision.\n\n![The earlier version v3 is retained in a history tray, while draft v4 awaits review.](https://ai-nterprise.com/assets/case-motion-final-desktop.webp)\n\n**v3: kept, not eligible for the affected new use.** Draft v4 requires review. No action executed.\n\n[Inspect the four moments](https://ai-nterprise.com/#case-study)\n\n05 / The ten principles\n\n## The principles behind the model.\n\n- Draft by default\n\n- Make promotion explicit\n\n- Let evidence travel\n\n- Treat provenance as necessary but insufficient\n\n- Keep the four dimensions separate\n\n- Scale scrutiny with consequence and persistence\n\n- Preserve independent judgment when warranted\n\n- Make challenge and removal first-class\n\n- Require assurance to change the system\n\n- Own the institutional learning loop\n\nOttaviano De Cicco · AI-nterprise: Governing What the Organization Knows · v1.0 · 3 September 2026.\n\n## Use and attribution\n\nThe articles interpret the framework; they do not claim that the complete architecture has been validated in enterprise field operation. Platform work is not evidence of improved outcomes.\n\nWhite paper: [CC BY 4.0 ↗](https://creativecommons.org/licenses/by/4.0/). The worksheet has its own CC BY-SA 4.0 licence.\n\n[Explore the vocabulary](https://ai-nterprise.com/glossary/)\n" }, { "path": "/glossary/", "title": "Vocabulary", "description": "Definitions and exact formulations from the AI-nterprise framework, with direct term links and the source DOI.", "content_status": "published", "h1": "Vocabulary", "canonical": "https://ai-nterprise.com/glossary/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/glossary.md", "html_sha256": "9640facf516f465a9fbfea7426fb203acce5369c803f181a8bffcc3edf35ddad", "markdown_sha256": "3e633c79456158e1e2e7ca278c5bd7e8dbc865f36e507d792f731934441588d4", "content_markdown": "Canonical: https://ai-nterprise.com/glossary/\nLanguage: en\nContent status: published\n\nGlossary · Public vocabulary\n\n# Vocabulary\n\nShort definitions from the public framework, with their source and limits attached.\n\nUse your browser’s Find command (Ctrl + F, or ⌘ + F) to search this page.\n\n[AI-nterprise](https://ai-nterprise.com/glossary/#ai-nterprise) [Organizational claim](https://ai-nterprise.com/glossary/#organizational-claim) [Epistemic standing](https://ai-nterprise.com/glossary/#epistemic-standing) [Persistence](https://ai-nterprise.com/glossary/#persistence) [Visibility](https://ai-nterprise.com/glossary/#visibility) [Action enablement](https://ai-nterprise.com/glossary/#action-enablement) [Knowledge plane](https://ai-nterprise.com/glossary/#knowledge-plane) [Authority plane](https://ai-nterprise.com/glossary/#authority-plane) [Assurance plane](https://ai-nterprise.com/glossary/#assurance-plane) [Learning loop](https://ai-nterprise.com/glossary/#learning-loop)\n\n**AI-nterprise**\n\nAn institutional architecture for coordinating humans, agents, models, claims, and rules so that an organization can know, decide, act, and learn without allowing authority, accountability, or memory to dissolve into the workflow. It is not a product category or model class.\n\n**Organizational claim**\n\nAn assertion, evaluation, decision, or rule that originates in a context, carries evidence and provenance, may acquire standing, may enable action, and may be reused as precedent.\n\n**Epistemic standing**\n\nThe decision that a claim is sufficiently warranted for a stated purpose, context, and level of consequence.\n\n**Persistence**\n\nThe decision to retain a claim for a defined period, in a defined version, under correction and disposal rules.\n\n**Visibility**\n\nThe decision governing which person or agent may retrieve which representation of a claim and its supporting context.\n\n**Action enablement**\n\nThe decision governing who or what may act on a claim, within which boundary and escalation path.\n\n**Knowledge plane**\n\nThe control plane for claims, evidence, versions, standing, and correction.\n\n**Authority plane**\n\nThe control plane for roles, permissions, thresholds, delegation, exceptions, and obligations.\n\n**Assurance plane**\n\nThe control plane that tests whether knowledge and authority controls work and whether the available evidence justifies continued reliance.\n\n**Learning loop**\n\nThe loop through which verified correction changes later retrieval, evaluation, policy, memory, or action. A finding that changes no future behavior has not yet produced organizational learning.\n\n## Exact formulations\n\n- “Capability is not authority.”\n\n- “Output is not knowledge.”\n\n- “Human presence is not meaningful human power.”\n\n- “Seeing is not believing. Remembering is not accepting. Accepting is not authorizing action.”\n\n- “The model may be rented; the learning must be yours.”\n\nOttaviano De Cicco, *AI-nterprise: Governing What the Organization Knows*, v1.0, 3 September 2026, [doi:10.5281/zenodo.22282125 ↗](https://doi.org/10.5281/zenodo.22282125).\n\n[Explore the framework](https://ai-nterprise.com/framework/)\n" }, { "path": "/", "title": "Knowledge, judgment, and authority", "description": "The book, open framework and practical resources for the human–agent enterprise, by Ottaviano De Cicco.", "content_status": "published", "h1": "Intelligence, with a mandate.", "canonical": "https://ai-nterprise.com/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/home.md", "html_sha256": "d549c42636018cb3b5f4eff28eb0092ce45f5ef96bb55044f4be17abe565462a", "markdown_sha256": "7daf208cb9849082582f0b2117ff638340fcfb98e87d0d21017a403dab84937c", "content_markdown": "Canonical: https://ai-nterprise.com/\nLanguage: en\nContent status: published\n\nA book about governing AI in the enterprise\n\nOttaviano De Cicco\n\n# Intelligence, *with a mandate.*\n\nA framework for governing what your organization knows, decides, and allows AI agents to do.\n\n[Explore the book](https://ai-nterprise.com/book/) [Explore the framework](https://ai-nterprise.com/framework/)\n\n**Diagram transcript:** The moment of consequence · A specific claim version reaches a gate. Knowledge and Authority converge on the same gate. One conditional route leads to an effect; another holds the proposed use for review. Assurance probes Knowledge, Authority and the gate and sends findings to the correction owner. No assurance connection enters the action route. · What is claimed · What is decided · What happens · v3 · Claim · a specific version · Four independent decisions: standing, persistence, visibility, action enablement · Gate · If enabled for this use · C2 · Effect within scope · Knowledge · Evidence · standing · Authority · Mandate · limits · Use held · Proposal for review · Assurance · independent checks · Owner · correction\n\n**The moment of consequence.** Knowledge and authority meet at use time. A held use leaves the stored claim version intact. Four independent decisions: standing · persistence · visibility · action enablement.\n\n### Inside the framework Four dimensions · three planes · three loops\n\nFour separate decisions\n\n**Standing**\n\nIs this claim warranted for this use?\n\n**Persistence**\n\nWhat remains eligible for use, and what history must be kept?\n\n**Visibility**\n\nWho may retrieve which representation and context?\n\n**Action enablement**\n\nWho may act, within which boundary?\n\nThree control planes\n\nKnowledge and Authority supply distinct conditions at the gate. Assurance independently tests both planes and the gate. Its findings go to the responsible owner for correction.\n\nThree coupled loops\n\n**Production** creates work and revisions. **Assurance** challenges and tests. **Learning** turns verified correction into a change in the next relevant case.\n\nThe square's four edge marks represent separate decisions; they are not four green approval lights. This is the book's reference model.\n\n01 / The framework in practice\n\n## One case, *four moments.*\n\nA customer's payment terms. New evidence changes whether an earlier assessment can support the next decision.\n\nChapter 9 · Concept illustration · No action executed\n\nCustomer payment terms · An illustrative case from Chapter 9\n\n### A claim supports a proposed change.\n\nAssessment v3 supports 60-day terms. No action has been executed.\n\n[Video: New accounts require review; a barrier closes before the proposed payment-term change; the earlier version remains in history](https://ai-nterprise.com/assets/case-motion-mobile.mp4)\n\n- **Evidence** Customer accounts\n\n- **Claim · v3** Liquidity assessment\n\n- **Use boundary · open** Barrier across the route\n\n- **Proposed effect · Net 60 days** 60-day payment terms\n\n1 · Basis\n\n2 · New evidence\n\n3 · Hold use\n\n4 · History\n\nPlay animation\n\nSelect a moment to inspect it.\nConcept illustration · No action executed.\n\nNew accounts require review. The affected use is held; the earlier basis remains in history.\n\n![AI-nterprise light paper and green cover](https://ai-nterprise.com/assets/book-cover.png)\n\nForthcoming · Q4 2026 · English edition\n\n[Discover the book](https://ai-nterprise.com/book/)\n\n02 / The argument\n\n## The organization *behind the intelligence.*\n\nAn agent produces an answer. Someone copies it into a report. Another system retrieves it. A team acts on it.\n\nKnowledge\n\n### Claim, evidence, and context.\n\nThe book examines how generated output becomes warranted knowledge.\n\n[Explore the vocabulary](https://ai-nterprise.com/glossary/#definitions-title)\n\nJudgment\n\n### Review with the power to challenge.\n\nWhy human judgment and confidence must be assessed alongside machine output.\n\n[Read the essays](https://ai-nterprise.com/articles/)\n\nAuthority\n\n### A mandate at the moment of use.\n\nA refund recommendation can be perfectly clear and still lack authority.\n\n[Try the action boundary](https://ai-nterprise.com/action-boundary/)\n\nStart here\n\n[Read the manifesto](https://ai-nterprise.com/articles/manifesto/) [Read the ten principles](https://ai-nterprise.com/framework/#ten-principles) [Use the Four Questions worksheet](https://ai-nterprise.com/tools/#four-questions)\n\n03 / The open framework\n\n## An open foundation. *A shared vocabulary.*\n\nThe open white paper, version 1.0, defines the model and its limits.\n\nWhite paper · Version 1.0 · 3 September 2026\n\n[Read on Zenodo](https://doi.org/10.5281/zenodo.22282125)\n\n![AI-nterprise cover: pale background, deep green AI and central node, dark title and pathways](https://ai-nterprise.com/assets/book-cover.png)\n\n04 / The book\n\n## AI-nterprise\n\nGoverning Knowledge, Judgment, and Authority in the Age of AI Agents\n\nFor the leader accountable for what an organization does with what it believes.\n\nForthcoming · Q4 2026 · English edition\n\n[Discover the book](https://ai-nterprise.com/book/) [Get the reader's resources](https://ai-nterprise.com/book/#resources)\n\n05 / Ideas\n\n## Questions worth *staying with.*\n\n[Explore all Ideas](https://ai-nterprise.com/articles/)\n\n[![Three glass screens and a wireframe cube becoming a solid green object](https://ai-nterprise.com/assets/editorial-manifesto.webp)\n\nManifesto · 4 September 2026\n\n### A Manifesto for the Human-Agent Enterprise\n\nThe public argument for governing what humans and agents may know, decide, remember, and do together.\n\nRead the essay](https://ai-nterprise.com/articles/manifesto/) [![A wireframe cube suspended above concentric green rings](https://ai-nterprise.com/assets/editorial-learning.webp)\n\nEssay · 15 July 2026\n\n### The Model Is Rented. The Learning Must Be Yours.\n\nWhy corrections and evaluations must become reusable organizational learning.\n\nRead the essay](https://ai-nterprise.com/articles/the-model-is-rented/) [![A solid green sculpture and its wireframe counterpart separated by glass](https://ai-nterprise.com/assets/editorial-judgment.webp)\n\nEssay · 12 June 2026\n\n### The Human Hallucination: The AI Failure Mode Nobody Is Measuring\n\nWhy judging the human–AI pair matters when fluent output changes a person's confidence.\n\nRead the essay](https://ai-nterprise.com/articles/the-human-hallucination/)\n\n06 / Practical resources\n\n## Take the questions *into your work.*\n\nFour Questions is the complete printable worksheet currently available for the reader.\n\nOne open download. No account or signup.\n\n[Explore the resources](https://ai-nterprise.com/tools/)\n\nPreview worksheets · Version and limitations shown with each file.\n" }, { "path": "/tools/", "title": "Resources", "description": "Download Four Questions Before an Agent’s Answer Becomes Action: instructions, a process map, a filled example, and an open licence.", "content_status": "published", "h1": "Put the questions to work.", "canonical": "https://ai-nterprise.com/tools/", "share_image": "https://ai-nterprise.com/assets/share-brand.png", "markdown_url": "https://ai-nterprise.com/agents/content/tools.md", "html_sha256": "e877f8b95d89116ee7e3bebbe4a9bad52e57b98e11658c6d7c1c4c503b67cccf", "markdown_sha256": "0deff7a102c043c117052f1c5277c472b923b6f133fb40e7519306a1fe8c1799", "content_markdown": "Canonical: https://ai-nterprise.com/tools/\nLanguage: en\nContent status: published\n\nResources · Open downloads\n\n# Put the questions to work.\n\nOne practical worksheet for examining what an organization may trust, remember, reveal, and allow to cause a consequence.\n\n## Four Questions Before an Agent's Answer Becomes Action\n\nUse this worksheet on one or two real processes in which an agent produces, transforms, recommends, or passes on a claim.\n\nThe result is a decision map: it shows where four different organizational decisions are explicit and where one generic “approve” is carrying too much weight.\n\nPDF · 2 pages · v0.1 preview · 2 October 2026\n\nInstructions, process map and filled example. Open download. No account or signup.\n\n[Download Four Questions](https://ai-nterprise.com/book/downloads/four-questions.pdf) [Open the PDF](https://ai-nterprise.com/book/downloads/four-questions.pdf)\n\n## Begin with one process\n\nWrite the current rule, its owner, and the evidence that the rule was applied. Mark what is explicit, partial, or missing. The sheet helps you name one boundary to clarify before the next comparable case.\n\nYour output is the completed map and one next decision. Keep it beside the workflow.\n\nThis is not a maturity score. A blank cell is a useful finding.\n\n[Check the vocabulary](https://ai-nterprise.com/glossary/)\n\n## Use, adapt, and attribute\n\nWorksheet content © 2026 Ottaviano De Cicco. Licensed under [Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ↗](https://creativecommons.org/licenses/by-sa/4.0/).\n\nCredit the author, link to the licence and indicate changes. Distributed adaptations must use the same or a compatible licence. The AI-nterprise name and trademark are excluded from this licence; this does not remove the attribution requirement.\n\nThis worksheet helps organize decisions and evidence. It does not certify compliance or validate the framework in enterprise field operation.\n\nOttaviano De Cicco, *AI-nterprise: Governing What the Organization Knows*, v1.0, 3 September 2026, [doi:10.5281/zenodo.22282125 ↗](https://doi.org/10.5281/zenodo.22282125).\n" } ], "records": [ { "title": "WebSigil: A Measurement Authority Architecture for AI Reliability", "publication_status": "published", "url": "https://doi.org/10.5281/zenodo.21177726", "description": "", "source_page": "/about/", "publication_date": "2026-07-03", "type": "technical-report", "doi": "10.5281/zenodo.21177726", "version": "1.0" }, { "title": "AI-nterprise: A Manifesto for the Human-Agent Enterprise", "publication_status": "published", "url": "https://www.linkedin.com/pulse/ai-nterprise-manifesto-human-agent-enterprise-ottaviano-de-cicco-jjmbe", "description": "The institutional question behind the AI-nterprise framework.", "source_page": "/articles/", "publication_date": "2026-09-04", "type": "manifesto" }, { "title": "The Model Is Rented. The Learning Must Be Yours.", "publication_status": "published", "url": "https://www.linkedin.com/pulse/model-rented-learning-must-yours-ottaviano-de-cicco-16xsf", "description": "How an organization retains what it learns while using rented models.", "source_page": "/articles/", "publication_date": "2026-07-15", "type": "essay" }, { "title": "The Human Hallucination: The AI Failure Mode Nobody Is Measuring", "publication_status": "published", "url": "https://www.linkedin.com/pulse/human-hallucination-ai-failure-mode-nobody-measuring-de-cicco-exjhf", "description": "Why human judgment and confidence must be assessed alongside machine output.", "source_page": "/articles/", "publication_date": "2026-06-12", "type": "essay" }, { "title": "AI-nterprise: Governing What the Organization Knows", "publication_status": "published", "url": "https://doi.org/10.5281/zenodo.22282125", "description": "", "source_page": "/framework/", "publication_date": "2026-09-03", "type": "white-paper", "doi": "10.5281/zenodo.22282125", "version": "1.0" } ] }