Over the last five weeks our team merged more than 260 changes to the SpecOps.AI platform and shipped them in about 100 production releases. Nearly all of it pointed at a single question: can a customer find value in the platform quickly, and trust what they find?
We came into September with the platform live and a lite launch behind us. The engineering was never the hard part. What kept me up was the risk of losing customers before we ever got them, because a first screen assumed they already knew what we know. That is where I see AI adoption fail again and again. The technology works, and people still walk away, because nobody designed for the person sitting alone with the product for the first time.
The last mile of AI is not code. It is customer empathy.
So we built for that person.
What's new: five focus areas
Onboarding that teaches itself. Available now. A guided path that follows what each customer has actually done, a Learn page, a first-run tour and screens that explain themselves. Sherpa drafts your Statement of Work and starting backlog from your own documents, and a person on your team approves them.
Governance built into development. Available now. Every change we ship passes our full review across quality, architecture, production readiness, security and AI governance before it merges. Customers can run GAUNTLET's rule scan on their own repositories on demand, with every verdict recorded. Rule packs written for each customer's own stack are next. A control only shows as met when evidence says so.
Smarter integrations. Available now. Link a repository and your first Sherpa score starts right away. A merged pull request finds and moves its own card on the board. A governed connection for your team's own AI tools is in final safety review.
Telemetry you can trust. Available now. Every card on the Reports page carries a 0 to 10 trust score with a tamper-evident history. Overwatch runs twice a day on every active project, comparing the work you track with what changed. The first part of Touch, our proprietary integration of environment sensors into the platform, is running in our own enclave.
The AI Gateway. Sherpa now, outside tools in final safety review. One governed door every AI action passes through before it touches your work: a short list of allowed actions, closed by default, with a record of who acted. Sherpa already works through it inside the platform. Outside AI tools join once the final safety checks pass.
A CMMC readiness path for defense suppliers
The Compliance Pack takes a defense supplier through the CMMC Level 1 self-assessment inside the platform: the practice checklist, the executive affirmation, a readiness plan whose approval creates the work, and a system security plan with versions and sign-off. SpecOps reviews the readiness scope before the customer signs it. The pack prepares teams; it does not certify them. The Level 2 practice library is visible as a clearly labelled draft while verification is completed.
Moving into our own enclave
On October 3, the SpecOps application moved into our own private enclave in our own data center. Customer data follows it there before general onboarding opens, and we will say when it has. Regulated customers need to know where their data lives and who controls the machine it runs on, and this is how we answer that. The same environment is how we will host and scale customer applications next.
Product availability
| Product | What it does | Availability |
|---|---|---|
| Front Door | Guided onboarding from signup to first value | Available now |
| Project and Scope | Statement of Work, backlog and board in one place; sign only what you read | Available now |
| Sherpa Command | Sherpa inside every project, grounded in your own data, with a visible trace | Available now |
| GAUNTLET | Rule scan on your own repositories, on demand, with every verdict recorded | Available now |
| Compliance Pack | CMMC Level 1 self-assessment readiness; Level 2 library as a labelled draft | Available now |
| Overwatch and Orchestrator | Monitoring and governance decisions based on your own evidence | Available now |
| Trust Ledger | A trust score on every report card, with a tamper-evident history | Available now |
| AI Gateway | One governed door for every AI action | Sherpa now; outside tools in final safety review |
| Connected AI tools | Your team's own AI tools, working the platform under the same rules | Final safety review |
| Touch | Environment sensors integrated into the platform | First part running in our own enclave |
| Customer data in the enclave | Your data in the same private environment as the application | Coming next |
| Hosted enclave | Customer applications hosted in our private environment | Coming next |
Five weeks, by the numbers
- More than 260 changes merged
- About 100 production releases
- A review report on nearly every change
October: show up and show out
September was preparation. October is when customers get their hands on it. This month we open live demos and onboarding, and we are choosing a small number of design partners to build alongside. My team will run every demo on the real product, with the customer driving as soon as they want the wheel, and we will turn what we hear into shipped changes they can see.
While that starts, we finish what is still in flight:
- Customer data moving into our private enclave
- Connected AI tools and the full AI Gateway, once final safety checks pass
- Touch sensor evidence flowing into each customer's controls
- Rule packs for each customer's own stack in GAUNTLET
- Hosting for customer applications in our private enclave
Some of this is still being finished, and I have told you which parts. Everything marked available is live today.
To see it, book a demo below.
About SpecOps.AI
SpecOps.AI is a governed AI execution and delivery platform: the operating layer that lets an organization use AI safely, with proof, on top of the systems it already runs. SpecOps enforces governance at the moment AI acts, records who did what, human or AI, works across any vendor or cloud, and is delivered by a team that builds the last mile and stays. Where the rules meet the work.
Media contact: Ron Griffin, CEO, SpecOps.AI, [email protected]
Figures cover September 1 to October 4, 2026, counted on October 4. "Available now" means live in production for customers. CMMC Level 1 is a self-assessment; SpecOps prepares teams and does not certify.