Command & control · Powered by EARL
Every action.
Under authority.
Bring governed intelligence to command and control. OVERWATCH checks the evidence, applies your rules, and determines which proposed actions are permitted. Every decision carries a signed record that can be independently checked.
The decision layer inside your operation
Check rules. Rank permitted options.
Signed records preserve the evidence, authority, and outcome for independent verification.
Watch OVERWATCH in action
See the decision.
Verify the record.
A recorded console demonstration of EARL's command-and-control runtime, presented as OVERWATCH. Follow declared rules, conflicting evidence, a referral for human authorization, and a signed record checked by a separate verifier.
This software demonstration uses synthetic inputs on a laptop, with no simulator or network required by the runtime. The authorization step is scripted in this recording; in the live console, a person supplies it. The video does not depict a live weapon engagement.
The operational challenge
More autonomy.
Clear accountability.
As systems move faster, leaders need to know that every consequential action has the evidence and permission it requires.
- 01What evidence supports this action?
- 02Who has authorized it?
- 03Do the operating rules permit it?
- 04Can we explain the decision afterward?
Built to integrate
Keep your systems.
Add governed decisions.
Your team keeps its sensors, tracking, fusion, command systems, and operator interfaces. OVERWATCH receives their evidence and candidate options, evaluates them against declared conditions, and returns a release or refusal for your integration to enforce.
Connect the operational picture.
Supply the readings and proposed actions your systems already produce. OVERWATCH evaluates that information at the decision boundary.
Define the authority.
Declare permissions and operating conditions as data. Inspect and change those rules without a software release. A preferred option cannot override a failed requirement.
Inspect the outcome.
See the evidence behind a release or the specific condition behind a refusal. Decisions are written to signed, linked records before the demonstration proceeds to action.
An instruction is not an authorization.
For a proposed lethal action, OVERWATCH refuses release without human authorization and refers the decision to a person. It cannot grant that authorization to itself. Refusal, referral, authorization, and any subsequent release are separately linked in the record.
What it gives your team
Confidence at the point of action.
Control that acts before release.
Apply evidence and authority requirements while a decision is being made, with an explicit outcome your systems can enforce.
Conflicting sensors mean a hold.
When your systems report sensor disagreement, OVERWATCH refuses the affected action and identifies the failed condition. Your fusion system resolves the conflict before the action is reconsidered.
Verification beyond our runtime.
A separate verifier checks the signed record chain and re-derives decisions from the recorded rules and readings. It runs without EARL or the engine binary.
One EARL. Multiple domains.
A common foundation for governed intelligence.
OVERWATCH brings EARL into command and control. WARRANT applies the same underlying engine to autonomous flight; TALOS applies it to robotics. Each integration supplies its own evidence, rules, and execution interfaces around a shared cognitive foundation.
For partners and integrators
Designed to be examined.
Where does OVERWATCH fit?
OVERWATCH sits between proposed actions and their execution within a command-and-control integration. Existing systems continue to produce the operational picture and candidate options. OVERWATCH evaluates their admissibility; it does not generate tracks, calculate weapon-target assignments, or replace fire control.
What can an evaluation team inspect?
Jou Labs can provide the signed demonstration records and a separate verifier, together with recorded software tests, decision replay, adversarial checks, and runtime observability. The verifier checks record integrity and re-derives decisions from the declared bounds. This is not independent confirmation that the source readings are true. The current C2 tests use synthetic, in-process inputs on a single host. Partners can evaluate the runtime with independently authored scenarios and their own integration conditions.
What does the integration need to own?
The integration supplies trustworthy readings, maintains their freshness, communicates authority changes explicitly, and enforces the returned disposition in its execution path. Stopping an authority feed is not the same as revoking an existing grant. Production signing-key custody is also an integration responsibility. These responsibilities are defined during technical integration.
Bring authority into
every decision.
Explore how OVERWATCH can fit inside your command-and-control architecture.
Discuss an integration