Use cases
Concrete patterns you can steal for internal decks: what breaks with HTML-only automation, and what improves when you publish policy-first JSON beside your human experience.
These are not exhaustive—they are anchors. Each pattern shares the same idea: declare what exists, what can happen, and what automation is allowed, instead of forcing every agent to reverse-engineer your UI.
Patterns by function
Customer support and success
Agents triage tickets, suggest next steps, and escalate with context. If they scrape the help center, a navigation or content tweak can silently change what they “see.” AOM lets you publish task graphs, allowed actions, and entities (case types, SLAs, product areas) so automation follows explicit structure—while your help UX stays free to evolve for humans.
B2B commerce and revenue operations
Quotes, approvals, renewals, and partner portals involve long-lived workflows and strict rules. DOM-based bots confuse cosmetic changes with business logic. Structured surfaces describe quotes, line items, approval states, and policy (who may submit, renew, or cancel) so agents and humans share the same source of truth—not parallel interpretations of a table layout.
Documentation and developer surfaces
Models answer questions about APIs, limits, pricing tiers, and migration paths. When answers depend on parsing docs HTML, small copy edits become “model drift.” Publishing entities and actions for limits, endpoints, and plans gives retrieval and agents stable anchors—reducing hallucinated parameters and wrong tier assumptions.
Internal tools and IT service management
Employee-facing portals for access requests, expense flows, or HR steps suffer the same fragility as customer sites—often worse, because fewer people notice when an internal script breaks. AOM-style JSON on those routes makes internal automation and copilots safer to roll out and easier to audit.
Regulated and high-stakes workflows
In finance, healthcare-adjacent, and public-sector contexts, “we blocked the bots” is rarely enough. Stakeholders want to know what is permitted, logged, and revocable. Pairing structured surfaces with clear automation policy supports those questions with artifacts you can point to—not ad hoc firewall rules alone.
How teams usually start
Most begin with one high-traffic or high-risk journey, validate JSON in CI and with public validators, then widen coverage. For vocabulary and schemas, rely on agentobjectmodel.org; for plugins, kits, and validation workflows, use aom.tools. When you are ready to align execs on outcomes and governance, the For businesses page frames ROI and adoption in one narrative.