Course-of-action (COA) analysis for constrained sustainment decisions.
JLAS-AI (Joint Logistics Artificial Intelligence System) helps staffs compare feasible courses of action for parts, maintenance and transport allocation, trace each to its readiness effect, and brief a decision package with its assumptions and sensitivities exposed.
What JLAS-AI does.
Constraint-aware COA generation
Separates hard physical, policy and mission constraints from commander preferences, then generates only feasible alternatives.
Resource-to-mission effect modeling
Traces Resource → Capability → Readiness → Mission Effect so staffs can show why one allocation outperforms another.
Decision-boundary analysis
Identifies the conditions under which the preferred COA would change, giving leaders a measure of how stable the recommendation is.
Human authority & decision provenance
Every package records inputs, options, rationale and the human disposition (approve, modify, reject). JLAS has no execution authority and fails closed when no feasible option exists.
Designed for how staffs actually decide.
- Operator-founded: built by a 23-year Army logistics and mobility veteran.
- Platform-agnostic: consumes approved data from existing systems of record through controlled adapters; doesn't replace them.
- Explainable by default: alternatives, margins and sensitivities are part of every output.
- Evidence-led: evaluations are scoped with success measures agreed before testing.
For contracting officers.
- Legal name
- Praevian Solutions, Inc.
- Location
- Orlando, Florida
- UEI
- Available on request
- CAGE code
- Available on request
- NAICS
- Available on request
- Business size
- Small business
- Point of contact
- Danny W. DeSilva, Founder & CEO · email
Define success before running the evaluation.
A useful evaluation is narrow enough to measure and realistic enough to matter.
Bound the decision
One mission workflow, a named decision authority and explicit resource constraints.
Agree the baseline
Document the current process, comparison method and representative approved data.
Predefine measures
Decision time, COA quality, constraint compliance, explainability and analyst workload.
Capture evidence
Inputs, candidate COAs, recommendation or abstention, rationale and human disposition.
Security grows with the data boundary.
The public demonstrator runs only on synthetic, unclassified data. Environments for sponsor evaluations and pilots are designed around the sponsor's data, hosting and accreditation requirements, with JLAS deployable into government-approved environments rather than requiring data to leave them.
Bring a bounded mission decision problem.
We'll structure a controlled evaluation around representative data, explicit constraints, measurable outcomes and human decision authority.