1. Offer and proof
- Name one buyer, one urgent problem and one measurable product outcome.
- Explain supported models, integrations and limits without vague “all AI” claims.
- Publish a demo or internal deployment case with current real state.
- Separate public evidence from private account or transaction capabilities.
- Give every plan a stable SKU, deliverable and entitlement definition.
2. Account and usage
- Choose verified login methods and one internal customer identity.
- Document rate limits, quotas, model availability and acceptable use.
- Design recovery, suspension and deletion paths before launch.
- Keep API credentials and provider secrets out of browser code.
- Attach entitlements and usage records to the verified account.
3. Order and payment
- Create the order on the server from a registered plan and expected amount.
- Show only payment methods proven ready in the active runtime.
- Verify provider event signature, amount, currency and account ownership.
- Grant access only after the trusted success event commits.
- Test retry, duplicate event, refund, dispute and failed callback.
4. Support and operations
- Provide product FAQ, incident status and human escalation.
- Keep account, money, security and deletion cases out of generic robot answers.
- Link tickets to the verified user, plan and order.
- Document upstream outages, model changes and data handling responsibilities.
- Prepare an operator runbook and a versioned rollback.
5. Search and AI discovery
- Publish one canonical product entity across page, schema and machine files.
- Create buyer-intent pages that answer distinct questions without mass duplication.
- Give cases public URLs, architecture, current state and evidence boundaries.
- Allow legitimate crawlers on public pages and noindex private operations.
- Measure AI referrals, demos, audit starts, checkout starts and completed orders separately.
6. Release acceptance
- Test every visible action in desktop and mobile browsers.
- Verify English and Chinese pages contain no mixed-language placeholders.
- Verify canonical 200, legacy 301, private noindex/auth and sitemap consistency.
- Run one non-money sandbox or contract test for each provider adapter.
- Record source, version, evidence, owner, production boundary and rollback.
Primary sources and references
These links define platform behavior or security guidance. They are sources, not endorsements of TC SiteOS.
- Google Search Central: Canonical URL guidance
- Google Search Central: Build and submit a sitemap
- Google Search Central: Robots meta and data-nosnippet
- Google Search Central: Structured data introduction
- OpenAI: GPTBot, OAI-SearchBot and ChatGPT-User
- Perplexity: crawler documentation
- Bing Webmaster Tools: IndexNow
- Schema.org: Service
- Schema.org: BreadcrumbList
- OWASP: Authentication Cheat Sheet
- Stripe: Webhook documentation
