Test & Control · AI Force Multiplication · Huntsville, AL
What I do, and what it costs.
First, who you are dealing with
Twenty three years across aerospace, defense and production companies, in design,
test, software and hardware. The programs changed, the job titles changed, and the
actual work almost never did.
I was the one sent in to sit down with a customer and work out what was really
wrong. Not to take an order. The problem a customer describes is rarely the problem
a customer has, and the part that mattered most was usually the part nobody had
asked me to look at. Find the real problem, say it out loud even when it is not what
anyone wanted to hear, propose the fix, then deliver it and stand behind it.
Do that for two decades and you notice you have been selling one thing the whole
time, under a lot of different names.
Discovery. Analysis. Delivery.
That is the company. Everything below is that same sequence at a different size, and
priced so you can buy as much or as little of it as your problem actually warrants.
Featured product
Hermes Foundational Build
ATCS's flagship AI product, and the platform every other engagement builds toward.
Every AI compliance story on the market today ends at the model endpoint. GovCloud Bedrock, Ask Sage, Azure OpenAI Government — all of them answer "is the model call allowed" and stop there. None of them answer for the agent making that call: its host, its file access, its tool calls, its logs, all of which touch the same data the model call does.
Hermes is ATCS's answer. An agentic operating system that knows what environment it's in, what data it's touching, and what that means, as a property of the runtime itself, not a policy someone has to remember to enforce. It's built and operated, not a whitepaper: the same dual-lane, classification-aware pattern behind every Hermes deployment, proven and extended with each one, sized to what your environment actually needs. You own what gets built. Nothing here is a subscription you can be locked into.
This isn't a one-time build. It's ATCS's core line of development — the product every engagement funds the next layer of.
A direct, hands-on mapping of what your organization actually runs today: existing tooling, identity and SSO, and exactly which of your systems a Hermes deployment would need to hook into. ATCS applies the same dual-lane architecture pattern already built and operated elsewhere against your actual requirements, not a generic template.
What you get
- A complete Agentic OS Foundational Plan: target-state architecture, lane structure, and integration points, specific to your environment
- Full tools-and-environment inventory
- Hooks mapping: which of your systems Hermes needs to integrate with, and how
- Lane determination: how many lanes you actually need, and what separates controlled work from open work in your own environment
How we'll know it worked
You receive a complete, environment-specific plan. It's yours outright: act on it internally, take it to another vendor, or bring it back to ATCS. The scoping fee buys the plan, not a commitment to anything after it.
One named, real workflow, deployed and running inside your own environment. A single lane, whichever the scoping determined is the priority. One real system hook integrated. Proves the pattern works on your actual infrastructure before any larger commitment.
What you get
- A working Hermes deployment, one lane, live in your environment
- One real system integration
- Your team understands the Hook and build pattern going forward
How we'll know it worked
The workflow runs on your infrastructure, and your team can operate it without me in the room.
The full dual-lane pattern live: open and controlled, both operating. The security classifier configured and tuned to your actual data-marking conventions. Multiple named workflows in daily use, not one.
What you get
- Both lanes live and tested, no shared endpoint, credential, or data path
- The classifier tuned to your marking conventions
- The tool-approval gate and audit logging operational
- Multiple workflows in daily use across your team
How we'll know it worked
Every workflow scoped for this tier is in daily use, and the audit log accounts for every tool call across both lanes.
Kubernetes/Docker high-availability deployment. Authentication integrated with your existing identity system. Every workflow identified during scoping, delivered. Your own engineers trained to operate and extend the platform independently.
What you get
- High-availability production deployment
- Identity integration with your existing system
- Every workflow identified during scoping, delivered
- Capability transfer: your engineers trained to run and extend it, no ongoing dependency on ATCS
How we'll know it worked
Your team ships a change to the platform without me touching the keyboard.
Scoping (2 wks) → Tier 1 (2 wks) → Tier 2 (5 – 7 wks) → Tier 3 (8 – 10 wks). Each tier is a real stopping point, not a funnel — you have something running and useful at every stage.
Tiers 1 through 3 are proven and available today. Beyond them, ATCS is building further: full label propagation through derived content (not just the original file an agent touches), continuous environment attestation instead of a one-time launch check, structured tool manifests in place of a binary approve/deny switch, and audit output mapped directly to NIST 800-171 control IDs. Every engagement funds the next layer of the platform.
Most consultants make you sit through a discovery call to find out their rate. Here
is the whole ladder, priced, with the acceptance criteria I expect to be held to.
The Briefing is discovery on its own. The Teardown is all three steps on one
workflow. Everything above it is the same three steps with more of your operation
in scope. Start anywhere. Most people start at the Teardown.
A working session with three to six of your leaders. I bring the North Alabama opportunity map, we walk it against your operation, and you keep a one-page ranked shortlist specific to your company.
What you get
- Ranked one-page opportunity shortlist for your operation
- Rough hour-cost estimate for the top three items
How we'll know it worked
You leave with at least three named workflows worth examining. If you don't, I've wasted your morning and I'll say so.
Most people start here
One workflow, not the whole company. I interview the two or three people who actually perform it, map where the hours go, and build one working automation against your real process.
What you get
- Working automation, running, handed over with source
- Written map of the workflow and where the hours actually go
- Hour-cost estimate in dollars at your wrap rate
How we'll know it worked
The automation runs on your inputs and produces output a member of your team accepts without rework. If it doesn't, you don't pay.
Five structured interviews, one half-day on site, and a scored map of every opportunity across engineering and business operations — ranked on impact, effort, and data sensitivity.
What you get
- Ranked opportunity roadmap with effort, impact and data-sensitivity scores
- A working proof-of-concept on the top-ranked item
- 12-month adoption sequence
- Build-versus-buy call on each opportunity
How we'll know it worked
Every opportunity is scored with a defensible number and the proof-of-concept demonstrably runs. No item is left as 'investigate further'.
Take the top opportunity to daily-use quality inside your environment. Acceptance criteria are written and agreed before any money is committed.
What you get
- Production-usable system running in your environment
- Runbook and operating documentation
- Training session for the team that will own it
- Handoff log covering every decision and known limitation
How we'll know it worked
The criteria we agreed in writing before the work started. They are specific, measurable, and I do not get to move them.
Embedded AI Champion
$2,500 – $4,500 / month
Ongoing · COO · CEO
Standing monthly capacity. I maintain what's built, extend it to the next workflow, train your people, and give your leadership a monthly readout.
What you get
- Maintenance and support of everything previously delivered
- One workflow extension or improvement per month
- Monthly written leadership readout
How we'll know it worked
Month to month, cancel with 30 days' notice. If it stops earning its keep you should end it, and I'll tell you when I think it has.
I teach your engineers to build this themselves. Tooling, patterns, prompt and evaluation discipline, and where the failure modes hide.
What you get
- Hands-on workshop series with your engineers
- Internal playbook written for your stack and your constraints
- One workflow built by your team with me reviewing
How we'll know it worked
Your engineers ship one workflow without me touching the keyboard.
The question that comes before all the others: can we even do this with our data? Tool selection, data-boundary rules, acceptable-use policy, and what CUI and CMMC actually permit.
What you get
- Written data-boundary policy for AI tooling
- Tool evaluation against your control requirements
- Acceptable-use policy your people will actually follow
- Shadow-AI exposure assessment
How we'll know it worked
Your ISSO can point at a written rule for any tool an engineer asks about.
Read the GovCloud authorization report — PDF, 10 pages
Not sure which one?
Start with the briefing. It costs nothing, it takes ninety minutes, and you'll
leave with a ranked list whether or not you ever hire me.
Book the briefing