SearchOS agents

The AI team that runs the discovery platform. Live agents are wired to real platform events today; others are scoped and activate as their surfaces ship. Click an agent for its activity.

Live

Onboarding

Application Onboarding

Drives the 5-minute onboarding: register an app, orchestrate discovery→generation→publish, get search live with no custom code.

Runs: When an app onboards

Live

Discovery

Schema Discovery

Introspects Prisma / OpenAPI / GraphQL / DB schemas into entities, fields, relationships and permission-scoping fields.

Runs: During onboarding

Live

Generation

Experience Generation

Zero-config: turns a discovery into a complete search profile — layouts, filters, facets, ranking, AI config, views, actions.

Runs: During onboarding

Live

Certification

Application Certification

Runs the certification checks (schema/permissions/relationships/runtime/bundle/search/AI/perf/a11y/security) and scores health.

Runs: On publish & sync

Live

Runtime

Runtime Health

Watches every app runtime: bundle/runtime versions, signature, compatibility, sync outcomes and rollbacks.

Runs: On each runtime sync

Live

Security

Permission & Security

Guards the ADR-0003 pre-filter and permission mapping; verifies signed bundles and red-teams for permission leaks.

Runs: On publish / continuous

Planned

Ranking

Relevance & Ranking

Tunes ranking weights and hybrid fusion, runs ranking A/Bs, and flags relevance regressions on the golden set.

Runs: On query feedback

Planned

Quality

Search Quality

Tracks nDCG / MRR / recall (targets), CTR, zero-result and reformulation rates; raises alarms on quality drops.

Runs: Continuous

Planned

Indexing

Indexing

Maintains each app's local index from published entity/permission/schema events; manages reindex and index health.

Runs: On data-change events

Planned

Copilot

AI Copilot

Grounded answers, summaries, comparisons and recommendations over results — cited, permission-safe, budget-bounded.

Runs: On AI query

Planned

Providers

Provider Optimizer

Benchmarks search/vector/graph/cache/AI providers on latency/cost/quality and recommends the best selection per profile.

Runs: Per deployment profile

Planned

Faceting

Faceting & Filters

Generates adaptive, context-aware facets and filters; learns which refinements matter for each entity type.

Runs: On query

Planned

Graph

Knowledge Graph

Builds and maintains the relationship / knowledge graph and powers permission-aware traversal and related-item discovery.

Runs: On relationship change

Planned

Personalization

Personalization

Learns per-user behaviour (recent, favourites, role) and personalizes ranking and result layout — permission-safe.

Runs: Per user

Planned

Analytics

Analytics & Insights

Aggregates search analytics into insights, dashboards and reports; feeds the Quality and Ranking agents.

Runs: Continuous