I'm Rich Bushnell. I built Ajutant.
I led the AI Centre of Excellence at a regulated UK energy business. The pattern I saw there, repeated in every regulated firm I talked to, was the same one: the people who understood AI couldn't get past governance, and the people who understood governance couldn't get to AI. Ajutant [AJ-oo-tant ] is what I left to build.
The pattern I kept seeing, until I decided to build for it.
The shape of the problem first showed itself at an Oracle conference in London. I was standing with our EVP of Infrastructure, who was adamant: no AI went near the business until he could be certain it would never surface data to someone who shouldn't see it. At the time I found it frustrating. I was leading our AI Centre of Excellence, ten teams were ready to move, and all of them were blocked behind that one decision. Looking back, he was completely right, and being right is what cost us a year.
I'd run the pilots myself. I built a small agent to answer a question as simple as how much I was allowed to spend on a hotel. The policy was sitting right there in a SharePoint document I could read with my own eyes. I could not get it to give me the answer. Worse, I had no way to prove the thing I actually needed to prove: that it would never surface something from that folder I didn't intend. In a regulated business, unprovable means unshippable.
Underneath the block, it wasn't stillness; it was pressure finding other outlets, and not only at the company I was in. Comparing notes across regulated firm after regulated firm, the shape was always the same: developers experimenting on one AI platform, a team somewhere quietly building its own thing on an open model because it wasn't going to wait, a leadership team wanting Copilot because the Microsoft bill was already paid. The people who went around governance shipped. The people who respected it looked incapable, not because the work was beyond them, but because doing it the right way had no platform to run on.
I left before I had the solution. I left because I'd finally understood the problem, and understood that I couldn't solve it from a chair inside one company. The answer was a platform that built the discipline into the build loop itself, so a regulated organisation could move fast and prove it was safe at the same time. Speed with discipline, not speed versus discipline. That's the thing I left to build.
Four positions, held strongly.
Every product reflects the worldview of the people who built it. Here are the positions Ajutant is built on. They aren't features; they're the reason the product looks the way it does.
Let's walk, not let's run.
The industry is pushing toward autonomous agents acting on their own initiative. Regulated organisations aren't ready for that, and shouldn't pretend to be. The right pattern is human-in-the-loop: AI does the work, a person stays accountable, the audit trail is real. That's what Ajutant is built for.
Your tenant, your data, your control.
Most AI vendors want your data in their cloud so they can run their service efficiently. That's a fine model for consumer AI; it's the wrong model for regulated organisations. The platform belongs in your tenant, under your control. We're suppliers, not custodians.
Speed and safety aren't opposites.
The usual choice is fast-and-uncontrolled (drop ChatGPT in front of your team) or slow-and-safe (an 18-month internal build). Both fail. The right answer is a build platform where the discipline is part of the loop, so shipping fast and shipping reversibly are the same operation.
The value is in your use cases, not in the platform.
A platform is infrastructure. Infrastructure is worth building only if there are use cases worth running on it. I'd rather help you ship ten real use cases that pay back than sell you a platform looking for problems.
Where I'm coming from.
Before Ajutant, two decades of running technology functions, the last ten of them inside regulated and operationally complex organisations. The specific roles, in reverse order:
What sits behind the titles: I've built and operated production systems that real businesses depend on, in environments where downtime, security failure, or regulatory breach have actual consequences. I'm not an AI specialist who's recently discovered enterprise IT. I'm an enterprise IT leader who watched AI become real over the last three years and decided to build the platform regulated organisations are going to need.
Honest about stage.
Ajutant is founder-led. I started it because the gap I saw was real, but I want to be plain about what that means in practice.
I'm not pretending to be a bigger company than I am. The site doesn't list customers we don't have. The security page is explicit about certifications we don't yet hold. Pricing is simple, and we'll walk you through it on the first call, no decoder ring required. Founder-led means I write the code, take the calls, and own the decisions; it also means I am personally accountable for every commitment we make.
The early customers are coming in through a Founder Programme designed for organisations who want to work closely with the people building the product. That's the right shape of relationship for this stage. When we have the operational maturity and the customer base to do something different, we will.
This is the part of the site where most companies put a logo wall, a list of advisors, and a careers page they haven't started hiring against. I've kept those out for the same reason I've kept everything else honest: I'd rather build trust slowly with the right buyers than win attention with the wrong signals.
Why "Ajutant".
Ajutant is an old word for an assistant or helper that was widely used until "adjutant" became the standard spelling. Pronounced the same way: AJ-oo-tant. We kept the older form because it reads as what it actually is: an assistant, not an agent. A helper, not a replacement.
The name is the position. Most AI vendors are racing toward "human-out-of-the-loop": autonomous agents that act on behalf of a human, who is then notified after the fact. That's the wrong direction for regulated organisations, and probably for most organisations.
What works is "human-in-the-loop": the AI does the work, the person stays accountable, the decision is made together. An assistant rather than an agent. In the loop rather than after the fact.
If you've read this far, let's talk.
Tell me what you're trying to do, what you've tried, and where it stalled. I'll be honest about whether Ajutant is the right answer for you. If it isn't, I'll say so.
Book a 30-minute discovery call