Situational intelligence. Built into the architecture.
Aurora is an intent-aware execution platform. Holmes uses deterministic and heuristic algorithms with configurable business rules. Deterministic intent inference that's fast, controllable, explainable, and private, without LLM cost or dependency. Replace static optimisation with systems that respond to human context.
Get started
Systems that adapt to intent
Build digital systems that understand human context and change behaviour in real time. Commerce is our first proving ground. The same infrastructure applies to marketplaces, CRMs, and internal tools wherever user state affects outcome.
Why Aurora
Platform capabilities
Holmes for inference. Aurora for execution. Structured data, event streams, workflow branching, and measurable outcomes. Infrastructure that understands the moment.
Holmes mission inference
Rules-based intent inference. Uses deterministic scoring, heuristics, and configurable business rules. Detects mission state, urgency, and context. Outputs confidence scores and execution directives. No generative AI. Deterministic intent inference that's fast, controllable, explainable, and private, without LLM cost or dependency. No chatbots, no surveys. Silent inference that changes system behaviour.
Execution directives
Holmes emits directives that adapt behaviour across systems. Mark storefront regions with data-holmes attributes; Holmes hides checkout extras, suppresses cross-sell, highlights payment, or expands discovery in real time. Fast checkout for urgent users. Coordinate storefront, inventory, and email based on inferred intent.
Predictive interface rendering
Holmes prefetches likely-next fragments (checkout summary, recommendations, catalogue list) before navigation. Content snaps in when users land - no extra round-trips. Optional SSE stream mode. Prefetch-hit analytics in Control Dashboard. Time-to-completion from first add-to-cart to payment.
Structured domain model
Tables, fields, and relationships that Holmes can interpret. Clean data is the foundation. Define your schema, add validation, bulk import. Real Postgres. Custom forms and cross-table views are built from your model with inline editing and AI assistance.
Event-native backbone
Every action is an event. Page views, cart updates, checkouts. Holmes consumes behavioural signals in real time. Domain events flow from state changes and trigger workflows. The source of truth for what Holmes observes.
Workflow branching
Workflows that adapt. Triggers on record events, webhooks, and mission state. Branch logic based on inferred context. Holmes-native templates provide launch-ready adaptive systems.
Ranking & discovery control
Adjust product order, recommendations, and UI based on mission. Urgent users see different results than explorers. Holmes switches optimisation targets in real time.
Observability & outcomes
See inferred mission distribution, confidence scores, and state transitions. Measure outcome deltas, friction reduction, and mission fulfilment. Holmes must be measurable.
Adapters & integrations
Plug Holmes into existing stacks. Event adapters and behaviour hooks for commerce, email, inventory. Adapters translate directives to system-specific actions. Aurora is the native adapter with zero friction.
Multi-tenant & security
Enterprise-ready isolation. Row-level security, audit logs, authentication. Production-grade from day one. The backbone must be secure.
API & storefront SDK
RESTful API, OpenAPI spec, and React/Next.js SDK. Connect storefronts, mobile apps, and external systems. Holmes integrates via a small script and receives inferred intent and suggested actions.
Hosted by default
No deployments. No servers. Sign up, design your data, wire Holmes, and ship. Event-native and inference run on the platform.
Public App
End-user app for customers and partners. Forms, views, and reports with App User role. Configure home sections, navigation, and CTAs. Full form parity - linked tables, filters, column visibility, row-click modals. No Studio access.