Receive
Adapters accept telemetry, redacted documents, images, maps, sensor batches, and route signals.
QURMMIUS INTELLIGENCE LAYER
One secure AI layer for every useful idea.
QI Pilot v1 · provider + local fallbackShared intelligence for telemetry, evidence, maps, images, sensors, and privacy-first workflows—so every Qurmmius product can stay focused while learning from the same trustworthy foundation.
The first milestone is live: take a structured input, route it to a product adapter, return an explainable result, and keep the next action in human hands.
Run a product skill
QI uses a protected server route when configured. Provider requests are redacted, rate-limited, correlated with a request ID, and kept out of browser logs. If the provider is unavailable, this page stays useful in local fallback mode.
Pilot mode is available without sign-in.
Structured result
Local analysis ready
Routed skill—
Confidence—
Signal—
Recommended next action—
Privacy mode—
Inputs / sources—
Provider mode sends only this redacted pilot input through the protected gateway. Local fallback sends nothing.
QI pilot controls
Active: server-side key handling · secret-pattern redaction · origin allowlist · request IDs · bounded recommendations · human-review flag · in-memory rate limit.
0 pilot runs · 0 provider tokens this month
LOCAL PILOT MEMORY
Every run leaves a small trail.
QI saves only a redacted preview, result signal, confidence, adapter, and request ID in this browser. Raw pilot input is not persisted.
Recent runs
- No pilot runs saved yet.
How the layer works.
Every product gets a narrow skill, while Qurmmius provides the shared controls around it.
Choose
A domain registry selects the right product adapter and the smallest model or local rule set that fits.
Explain
The adapter returns a structured signal, confidence, source trail, and a bounded recommendation.
Review
People approve consequential actions, while policy checks and an audit record stay attached.
PRODUCT ADAPTERS
Focused intelligence, shared guardrails.
Telemetry & coaching PilotApex Academia turns lap and setup signals into explainable review cues. No medical or racing-outcome guarantees.
Privacy & handoffs PilotContextKey Shield reviews origin, credential, schema, and least-privilege risks without reproducing secrets.
Evidence & schemas PilotClearGrant maps verified, self-reported, stale, and missing evidence to a reviewer-ready next step.
Maps, worlds & form StagedChronosMap × EraWeaver and Fluid Fluidity preserve provenance, uncertainty, and study context.
Signals & stewardship StagedEcoEcho and GridPulse aggregate useful trends without exposing individual histories.
Schema validation StagedSchemaForge flags mismatches and possible trackers using synthetic or redacted payloads.
NEXT DEFINED MILESTONES
Build the layer in proof-sized steps.
Each milestone turns one part of the pilot into a reviewable production capability. The next gate is secure access and durable audit history—not more surface area.
QI Pilot v1
The shared request path is working end to end.
- Provider gateway with local fallback
- Three named pilot adapters
- Redaction, rate limit, request ID, usage metadata
- Local redacted run history
Trusted workspace
Make QI safe for a small invited team.
- Real login and tenant isolation
- Durable audit records and retention controls
- Telemetry and evidence ingest connectors
- 20–50 golden evaluations per pilot adapter
Measured platform
Operate QI as a paid, observable service.
- Per-tenant usage metering and Stripe plans
- Admin cost, latency, and quality dashboard
- Approval workflow for consequential actions
- Staged adapters and tool calls for the remaining products
Connector contracts, ready for implementation.
These narrow interfaces keep future integrations reviewable and prevent product pages from handling provider credentials.
Telemetry ingest/api/qi/ingest/telemetry · normalize laps and quality flags
Evidence ingest/api/qi/ingest/evidence · preserve dates and provenance
Source search/api/qi/search/sources · return citations and uncertainty
Action verification/api/qi/verify/action · require approver and audit ID
Planned AI-layer pilots.
Target pricing for validating the shared layer with a small team. These are not live subscriptions.
$49/moStarter · 1 product skill
$149/moStudio · 3 product skills
$399/moPlatform · up to 10 skills
Production rollout adds authenticated workspaces, durable audit storage, provider credentials, usage limits, evaluations, and human-approval policies before real customer data is handled.