XTAM Public Sector

You bring the mission.
We build the AI.

XTAM Public Sector builds AI systems for primes, defense contractors and research partners. Your engineers and scientists know the mission. We design and build the AI that works beside them. If your bid or program is missing the AI piece, that's the gap we fill.

Who we build for

Primes, defense contractors and research partners.

Primes

Add an AI builder to your bid or program team.

We work next to your own engineers and subject-matter experts. You lead the program. We're accountable for the AI.

How we teamWays to team with us

Defense contractors

Put AI to work on your own programs and engineering data.

Built inside your program's constraints. Start with a two-week AI Assurance assessment or a one-month POC phase. A full build takes six to nine months. Nothing longer than nine.

IR&DTurn this year's IR&D into AI you own.

Research partners

You bring the science. We build the data pipelines and models.

That's how we work with a research university, our teaming partner and research collaborator.

ResearchCritical minerals data pipeline

On what government now expects from contractors

“Government employees are expecting the contractors to provide them with AI first solutions. No longer AI bolted on solutions.”

Dr. Ryan Nguyen, Artificial Intelligence (AI) Capability Lead, Arcfield

Capability statement

XTAM Public Sector LLC

Legal name: XTAM Public Sector LLC. Small Business. SAM.gov registered and active. One-page PDF on request from govcon@xtam.ai.

NAICS

541512 · 541511 · 541715 · 541519 · 518210

PSC

DA01 · DA10 · R425

UEI · CAGE

TGZ2EXSVJWG3 · 190Z5

Business type

Small Business (SB) · XTAM Public Sector LLC · Miami, FL

Government contact

govcon@xtam.ai

NAICS

541512 Computer Systems Design Services
541511 Custom Computer Programming Services
541715 R&D: AI / Machine Learning
541519 Other Computer Related Services
518210 Computing Infrastructure & Data Processing

PSC

DA01 Business Application Development Support (incl. DevSecOps)
DA10 Application Development Software as a Service
R425 Engineering & Technical Support Services

Procurement

Open market
Micro-purchase (GPC)
Subcontracting / teaming
SBIR teaming
Rate card on request

Capabilities

What we build, by mission domain.

Engineering knowledge graphs

Knowledge graphs built from engineering documents and drawings, where every element traces back to its source.

Engineering knowledge graph · Codebase discovery assessment

Tracking and space domain awareness

World models learn how the real world behaves and forecast the next move of people, drones and orbital objects. In Space Domain Awareness (SDA), that forecast can save lives.

Mantis · Space Mantis

Scientific and environmental data

Domain expertise and massive scientific datasets don't produce insight on their own. We build the pipelines and models that do, working alongside your scientists.

Seafloor mapping · Critical minerals data pipeline

Technical document intelligence

Answers from hundreds to thousands of unsorted documents. Each answer traced to its source and built with intent assurance. Designed for a scale where standard RAG breaks.

Standards answers · Underwriting audit trail

Secure deployment

AI designed to run inside your boundary, including air-gapped. LLM-light: mostly deterministic, with constrained and checked LLM steps.

Engineering knowledge graph · Codebase discovery assessment

Computer vision and geospatial

Vision models that take the first pass on imagery and survey data. Your experts review findings, not raw data.

Seafloor mapping · Mantis

Codebase assessment

Know exactly what you own or will own. We build a complete map of your software estate from its code and architecture.

Codebase discovery assessment

Custom and private models

Models trained for one job that run where you choose, down to an air-gapped network.

Threat classifier · A private coding model

High-trust automation

Ask the same question a thousand times. Get the same answer every time. That's intent assurance. Our assistants answer from your sources and check their own work.

Ask AI · Answer-check classifier

Security and data handling

Five questions to ask about any AI on your program.

We design every system against these five questions, the AI Assurance layers. None of them is ever finished. Ask them of us, and of anyone else bidding the work.

01

Sovereignty

Are we building our own AI, or renting someone else's?

We can build without receiving your sensitive data. With a defense prime, we built an intelligent knowledge graph over engineering documentation, without receiving CUI. We built it; the prime tested it. Our codebase assessment reads code and architecture, never the data your systems hold.

02

Security

What can our AI reach, and what could it open up?

LLM-light: mostly deterministic, with constrained and checked LLM steps. We start from a secure foundation and box the generative model so it can't open a hole. Our systems are designed to run where your data lives, on open-weight or frontier models, in enclaves, or air-gapped.

“Cost and security concerns are top of mind for me. Just because you can with an open weight model, you could open up Pandora’s Box if you’re not careful.”

Jason Ordway, CTO, Olo

03

Governance

Who checked what our AI produced, and what are we putting into AI tools?

Every output traces to its source document and page. When sources conflict, we keep both and show where each came from, so your experts can resolve it. Low-confidence output goes to a person before it moves on.

04

Intent

Does our AI give the answer we meant, the same way every time?

We move decisions from open-ended generation to rules and classifiers, and check that identical runs give identical results. No model is right every time, so we design for answers that are consistent, visible and correctable.

05

Cost

What are we spending on AI, and what is it returning?

We cap tokens, route each task to the right model, and build small purpose-built models where they cut compute. We size the system to the hardware your program actually has.

Posture

CMMC Level 2 (self-assessed)
NIST SP 800-171 implemented
FISMA Moderate control alignment
SOC 2 operational experience
Designed for GovCloud, Azure Gov and air-gapped operation

R&D

AI is more than LLMs.

Internally funded R&D, listed to show where we invest, not as current offerings.

World model

Mantis

Mantis is our own world model, built with our own R&D. It predicts where a person or drone is heading, so a system can flag an approach to a restricted domain while there's still time to act.

CapabilityPredictive world models

Space domain awareness

Space Mantis

Space Mantis applies the same idea to orbit: telling routine station-keeping from rendezvous and proximity maneuvers, using public behavior data.

CapabilityPredictive world models

Selected work

Problems we work on, starting with engineering documentation for defense programs.

Defense

Engineering knowledge graph for a defense prime

A defense program's design knowledge lives in thousands of pages of engineering documentation. Generic chat tools lose the structure, and they don't hold up at program scale. With a defense prime, we built an intelligent knowledge graph over engineering documentation, without receiving CUI. Engineers query it from requirements down, and every answer traces back to its source. The target architecture is designed for GovCloud, Azure Gov and air-gapped operation.

ResultWe built it; the prime tested it.

Geoscience

Seafloor mapping

Scarce experts hand-label technical data, which caps how many projects a firm can bid on. We built neural networks, with no LLMs, to take the first pass on seismic profiles, so geologists review findings, not raw data.

ResultWork that took six or seven months per project dropped to about two weeks, at about 70% labeling accuracy.

Research university

Critical minerals data pipeline

Researchers hold soil-microbe and satellite data but lack the engineering to model it. As a research university's teaming partner, we would build the data pipelines and machine learning that let its scientists test what that data can reveal about critical minerals.

RoleTeaming partner and research collaborator.

Education

Threat classifier

Across tens of millions of messages, a general-purpose model misses too many rare, high-stakes threats. We built a model trained for that one job.

Result89% of threats flagged against 53% for a general-purpose model on our own benchmark, at about 50 times lower cost.

For defense and government contractors

Turn this year's IR&D into AI you own.

An AI platform license is capability you rent, and any other bidder can rent the same one. Put your IR&D into a system your company owns, runs and keeps improving. Your fiscal year sets the clock. Tell us when yours ends, and we'll plan a POC phase that fits.

Bring us the gap.

Send us a solicitation or a capability gap. We'll tell you what we would build and where we'd fit on your team. We reply within one business day.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Or email govcon@xtam.ai

How we team: teaming routes, codes and posture. Commercial work: hello@xtam.ai