Executive page: "How AI actually lands in an enterprise" — one page a chairman or GM can be forwarded (policy → ontology → enforced trail, three-step pilot) #369

Why

Maintainer instruction (2026-11-04), verbatim: 「高优先级的立卡」. That instruction covers the high-priority findings of a read-only comparison between the site and a leadership talk on enterprise AI adoption.

The site already explains ontology, governance and legacy integration well. But it speaks to IT and developers. There is no page written for the business decision-maker: a chairman, GM or CIO who has been forwarded a link and asks "how would this land in my company?"

• The /use-cases/ work (决策：官网面向业务用户——是否新增按业务职能组织的「使用场景」入口（参考 Claudeforce） #359, PR feat(use-cases): /use-cases/ by business team, home scenario cards, Why nav item #368) answers "what can my team do with it", organised by business function.
• This card answers a different question: "how do we adopt AI as an organisation, safely, and where do we start?"

What the page says (outline)

1. Answer first. AI pilots stall less on model quality than on the organisation:
   • rules that live in people's heads;
   • legacy systems the AI cannot reach;
   • no trail when it acts.
2. The concept to coin: 「制度写得越清楚，AI 越能放手干」 / "The more clearly the rules are written, the further AI can be trusted to act." Ground it in a single example.
3. One rule shown in three states. This is the proof artifact. Use a generic travel-expense rule, not a customer's:
   • (a) the policy text;
   • (b) the same rule written into the ontology (object, field, validation, approval);
   • (c) the runtime enforcing it, with an audit row. Include a read-only identity being refused.
4. Where AI connects to what you already run. Link the existing legacy-modernisation cluster. State plainly what is early: federation is read-only by default (src/lib/positioning.ts).
5. Three-step pilot.
   • Pick one process with written rules.
   • Write it into the ontology.
   • Measure four numbers: hours saved, cycle time, error rate, share of items auto-closed within authority.
   • Link /customer-stories/ pilot criteria.
6. What it does not solve (AGENTS.md rule 7):
   • unwritten policy;
   • processes that need judgement nobody has written down;
   • integrations that do not exist yet.

Constraints

• No customer names, logos or customer numbers. customer-stories.ts policy stands.
• No third-party statistics on the page. Trend figures age fast and are mostly self-reported.
• No model-vs-model comparisons. Do not lead with open source. Keep the tone humble.
• Every product claim must match a shipped mechanism page (/permissions/, /approvals/, /trust-center/).
  • Export of the ontology from Cloud/Enterprise is planned, per positioning.ts E1. Do not imply it ships.
  • Any "rollback" or "auto-close within authority" wording must cite a shipped mechanism, or be left out.
• Anything that says what ObjectStack is reads positioning[locale].t2. It is not re-authored.
• en + zh-Hans sources (other locales fall back to en for marketing pages).
• Browser-verify at the six AGENTS.md viewports.

Open for the maintainer: placement

The placement must be ruled before dispatch.

Seat recommendation: a sixth topic cluster in src/lib/clusters.ts (slug enterprise-ai-adoption).

Reasons:

• A cluster is the existing shape for a forwardable topic hub, with linked articles.
• It adds no nav item. 决策：官网面向业务用户——是否新增按业务职能组织的「使用场景」入口（参考 Claudeforce） #359 measured that a sixth top-level item overlaps at 1100 px in de/es/fr.
• It does not compete with /use-cases/. That page is function-by-function; this one is organisation-level.

Alternatives:

• A section on /use-cases/. It would dilute that page's by-function shape.
• A new marketing page under 「为什么选 ObjectOS」. It would be a fifth item there, so the header width needs re-measuring.

Per the maintainer's instruction 「都不要用fable」, this recommendation was not run through a fable evaluation.

Related

• 决策：官网面向业务用户——是否新增按业务职能组织的「使用场景」入口（参考 Claudeforce） #359 / feat(use-cases): /use-cases/ by business team, home scenario cards, Why nav item #368: /use-cases/ by function. This page links to it, and does not duplicate it.
• The scenario-readiness card and the trust-center five-gates card were filed in the same batch.
• objectstack-ai/cloud#3567: a scenario test card that measures whether the cloud agents support these claims. The page's capability wording should follow its results.