We Didn't Add AI to Our Services. We Rebuilt the Business Around Outcomes.

Published Oct 5, 2026 · on LinkedIn ↗

Part of the Operate with Brian series.

For years, BSC Analytics looked like a lot of strong cloud shops. We ran managed services. We ran DevOps and platform work: migrations, modernization, automation, and the operational discipline that keeps cloud estates reliable at scale. And we staffed it a certain way on purpose: senior engineers only. People who could do the work and explain the why to a business stakeholder without needing a translator.

That model worked. Until it didn't. Managed services got thin. Margins compressed. Competition piled in. We didn't romanticize it. We phased it out. DevOps services stayed. And we weren't an hours shop. We sold story points, and we guaranteed them. Clients got value and velocity every sprint because the commitment was real. That made us one of the stronger services providers in a market full of teams that sold time and hoped for throughput. But even a strong story-point model has a ceiling. It tells you the team is delivering. It doesn't always tell a Fortune 50 buyer whether the business needle moved.

Then AI showed up as a real disruptor, not a slide in a strategy deck. The question wasn't how do we sprinkle copilots on billable work. It was harder: what happens to a firm of senior engineers when software can write a lot of what they used to type? Our answer was deliberate. We moved senior talent toward thought leadership and strategy, the work that decides what should be built and why it matters. We used AI to go faster, lower client cost, and reduce risk in regulated environments, with humans in the loop where accountability actually lives. Not as theater. As design. And part of the pivot was building product.

We didn't go hire a separate product army. We leveraged the engineering talent we already had, the actual experts on the execution, upskilled them, and built the product with them. We only hire people who like to just figure it out. And figuring it out they did. TACE was our entry: a way for clients to describe what they needed in plain language and get production-ready infrastructure and delivery patterns without needing that specialty on their own bench. Ariad was the continued evolution of that idea. It is an AI orchestration platform whose agents do the work, coordinated with people, to get the outcome done with governance built in. The framing matters. This isn't software-as-a-service bolted onto a services firm. It's services as software: the judgment, the delivery pattern, and the accountability, encoded so outcomes don't depend on who happens to be staffed that week.

That forced the next shift, and it was bigger than tools. We flipped from guaranteed story points and sprint velocity to outcome guarantees. Engineers sit with the business and define measurable outcomes, the kind you can argue about in a steering committee without hiding behind a burndown chart. Ariad is the get-it-done layer. Its agents execute the work. It defines done and success criteria, manages human and digital resources together, keeps visibility honest, and produces the audit packages regulated environments actually need. Under the hood, that includes modules for MCPs, governance over other AI systems, contracts that limit agents to the work they were asked to do, and clarity, per outcome, on what AI owns versus what a human must own. The point isn't autonomy for its own sake. The point is enforceable delivery.

Here's what surprised people outside the building: clients didn't push back. We've had real success with Fortune 50 companies. When you stop selling activity and start guaranteeing outcomes, with the controls to prove it, buyers who live in risk and audit don't flinch. They lean in. The flagship proof is a Fortune 50 replatforming and mainframe migration program. Traditional economics sat around roughly a million dollars per application. With this model, we drove that toward about a hundred thousand dollars per application, roughly a quarter of the time, at higher quality. Same class of work. Different operating system for how the work gets done.

None of that lands if the team can't change how it sees itself. Our people had to unlearn something sacred in engineering culture: the idea that value equals hands on the keyboard. Trusting AI without becoming a rubber stamp is a skill. The key was getting precise about what engineers validate, and why. Judgment moved upstream. Execution got faster. Accountability didn't disappear. It got sharper.

That's the transformation in one line, and it's the line I keep coming back to with clients and with our own leaders: AI doesn't create value by being used. It creates value when someone decides what matters first. Operate with Brian is about that decision, and building companies that can keep making it under pressure.