Where does the data go?
Inference, retrieval, and storage all run inside the isolated BlueBoots account
we provision for your agency. Prompts and case details are not shared with third-party
model providers, are not used for training, and do not cross account boundaries.
For agencies with stricter requirements we deploy a fully self-hosted open model on
dedicated infrastructure for your account alone — same product surface, harder isolation guarantees.
Is this CJIS compliant?
Compliance is a path, not a checkbox. The architecture is CJIS-aware from day one:
audit logs for every query, retrieval, and answer; encryption at rest and in transit
using FIPS-validated modules; controlled access with screened personnel.
Authority-to-operate (ATO) for any specific agency depends on your state CJIS Systems
Agency — in Minnesota that's the BCA. We help with the documentation and
walk through the process during deployment.
What jurisdictions are covered?
The pilot covers federal code, Minnesota statutes, and the relevant county and
city codes for the pilot agency. The architecture is jurisdiction-tagged at the
section level, so adding any new state, county, or city is a configuration change —
we ingest the canonical sources, the model retrieves with jurisdiction awareness,
and citations link back to the official source.
How does it know it's not making things up?
Every answer is grounded in retrieved excerpts from your jurisdictions' actual code.
The model is instructed to answer only from those excerpts and to say so explicitly when
the answer isn't in them. Citations point to specific ordinances and sections that
supervisors can verify with one click. Hallucinated statutes have nowhere to hide.
Will officers actually use it?
They're already using something. The shadow workflow we replace — ChatGPT on
personal phones — is in active use in most departments today. Our job is to
be at least as fast (sub-3-second answers), at least as easy (mobile-first,
one-handed UI), and dramatically more trustworthy. The honest test is the demo:
we deploy, you watch what happens for 30 days.
What does it cost?
We're building pricing alongside the pilots. The model that fits small departments
best is per-seat with a small annual minimum, billed against your agency directly.
Larger jurisdictions or county sheriffs typically prefer flat per-agency tiers.
Either way: no surprise per-token charges, no overage cliffs.
Concrete numbers come on the demo call.
Does it integrate with our RMS or CAD?
Not in v1. We start as a side-by-side reference tool that doesn't touch your RMS
or CAD. Once we have real usage data from your officers we add the integrations
that matter most. The most common requests we expect: pull a charge code lookup
directly into the report writer, attach citations to a CAD incident.
What if state or local code changes?
We refresh state statutes on the publishing cadence (Minnesota refreshes annually,
with session-publish updates), county ordinances monthly with change detection,
and city codes monthly. Federal code refreshes quarterly. If a section changes
between refreshes, we re-ingest on demand — no waiting for the next scheduled run.
Can supervisors see what officers are asking?
Yes — with role-aware controls. Admins and supervisors see audit logs of every query,
retrieval, and answer for the officers under them. Officers see their own history.
Nothing is anonymous, but nothing is broadcast either. The same audit trail satisfies
the CJIS audit and accountability requirements.
Who builds this?
BlueBootsAI is built by Imagination Guild LLC. Our pilot agency is a municipal
police department in Minnesota. We talk to Chiefs, watch officers use the
product, and ship improvements weekly during the pilot phase.