Investigation speed. Exam-ready proof.
Pilot
A 21-day plan to prove case prep speed and exam-ready exports.
How we're different
- Evidence collected automatically.
- Narratives cite every claim.
- Humans approve high-risk actions.
Trust rule: AI drafts, humans decide for sensitive outcomes.
21-day pilot timeline
Week 0Access + baseline
Connect one alert source, map required evidence types, and measure today’s time-per-case.
Week 1Evidence + graph
Ingest evidence objects, build the evidence graph, and enforce provenance on every artifact.
Week 2Narratives + QA
Draft cited narratives and run citation QA checks; track return loops and rework rate.
Week 3Gates + exports
Turn on supervisor gates and export exam bundles with audit stamps and integrity proofs.
Success criteria
Metrics we measure
- Time per case (median)
- % time spent gathering evidence
- Rework rate (supervisor returns)
- Citation compliance rate
- Backlog aging by queue
What you get
- One playbook tuned to an alert type
- Evidence graph + narrative templates
- Supervisor workflow and rationale capture
- Exportable audit-ready case bundle
- Pilot readout with before/after metrics (hypothetical model)
Next step
Ready to scope a pilot?
We’ll start with one alert type and measure outcomes from week one.
Request pilot plan