Products / Sheridan Labs

What we’re building.

Two owned software products for exacting technical work. They serve different markets, but both are built around authoritative sources and explicit boundaries.

FaultNavControlTrail
SL

Owned products

Built for work where correctness matters

The company site names both products without making availability claims. Each product will speak to its own market on its own property.

Automotive service

FaultNav

Guided diagnostics for technicians. It walks a tech through the manufacturer’s own diagnostic procedure, step by step, hands-free—and never tells them what the fault is. The tests do that.

Design boundary

The AI orders the path. Licensed manufacturer data defines what test results mean.

Built voice-first, so a technician can work a procedure with both hands in the engine bay.

Defense contracting

ControlTrail

CMMC readiness for contractors in the defense industrial base. It tracks evidence against all 110 NIST SP 800-171 controls, so a contractor knows exactly where they stand before an assessor asks.

Design boundary

The operational system of record underneath a customer’s SSP—not a certifying body.

The operational system of record underneath a customer’s SSP—not a certifying body.

Company structure

Separate markets. Shared standards.

FaultNav serves automotive technicians. ControlTrail serves contractors in the Defense Industrial Base. Their audiences do not overlap, so each product gets its own language while Sheridan Labs remains the company behind both.

Automotive service

FaultNav

Built voice-first, so a technician can work a procedure with both hands in the engine bay.

Defense contracting

ControlTrail

The operational system of record underneath a customer’s SSP—not a certifying body.

Shared engineering discipline

The same constraints follow both products

We don’t let software guess.

Our AI decides what to look at next. It never decides what the answer is. When FaultNav explains a test result, those are the manufacturer’s words—quoted exactly and carrying their citation—not a model’s paraphrase.

We’d rather be unhelpful than wrong.

When our software can’t be certain, it says so and hands you the source it was working from. It never fills the gap with something plausible.

Important rules are enforced by the build.

ControlTrail isolates each customer’s data in three independent layers. FaultNav’s tests fail the build if the AI is ever handed content it is not permitted to see.