Law firms have legal engineers. Go-to-market teams have GTM engineers. Palantir built a whole category around the forward deployed engineer. Now the same pattern is arriving in accountancy, and it may be the most important new job title the profession has seen in decades. 

At Ravical, the organisational AI company building for full-service accounting and advisory firms, we’re hiring for a role we call the Accountancy AI Engineer. To understand what the job actually involves, I sat down with Kimberly Vlietinck, Lead Product Manager at Ravical. Originally trained as a tax lawyer, she moved into software eight years ago and has spent the past year building Ravical from the ground up: sitting next to specialists inside firms across Europe, building AI workflows and agent systems with them, and watching what happens after she leaves the room. 

First things first: what is an accountancy AI engineer? You've been doing a version of this job before it had a name. How do you describe it? 

"It's someone who knows accounting deeply, builds AI workflows and agent systems hands-on, and can teach others to do the same. Those three things sound simple until you try to find them in one person. 

The core of the job is co-building. You sit next to a specialist at a firm, a tax adviser, a legal expert, an HR professional. You take a piece of work they do regularly, and together you turn it into a workflow the whole firm can run. It can be a VAT return review, a year-end file, an advisory memo. But because what we’re building is agentic, there’s so much that can be built into a workflow now: systems that loop, check, decide and act across a whole client book."  

So it's an accountant who builds, rather than an engineer who learned how an accountancy practice works? 

"In our case, yes, and that's deliberate. A software engineer can learn what a workflow needs technically. What they can't shortcut is knowing why a partner will reject a memo that's ninety percent right, or which step in a year-end close is sacred and which one only exists because someone set it up that way in 2011. The judgment and deep understanding comes from having done the work yourself. The building part is far more learnable than the domain part. That's what changed with modern AI tools: you no longer need to write code to build something real." 

Hiring for this role is helping Ravical’s accounting and tax firm clients, but why is it such a hot topic now? Why does this job exist in 2026 and not five years ago? 

"When you think about AI, five years ago feels prehistoric almost. We’ve seen the technology became capable of doing serious work, drafting advice, checking files, spotting what a regulatory change means for a specific client or for a firm’s complete portfolio. And at the same time, the pressure on firms became structural. Compliance margins are starting to compress as automation absorbs that work. Experienced people are hard to find and harder to keep. And clients are using the same AI tools their accountants use, so they expect more in return.

This technology is genuinely new. Firms aren’t just adopting a new tool. They’re navigating territory that didn’t exist two or three years ago. There’s no playbook and things are moving fast. Part of what this role is about is helping firms make sense of that shift as it’s happening, not just deploying a new tool. The accountancy AI engineer goes in as a guide as much as a builder." 

Couldn't firms just buy software and figure it out? 

"Some will. But there's a real gap between having a broad LLM and a capable partner that helps you reimagine how your work actually gets done.

Accountants do know how to decompose work. If this, then that, then this. The old decision trees were built exactly that way. At some point you'd hit a wall: too many rules, too many exceptions, too many layers stacked on top of each other. You couldn't maintain it. So the logic ended up living in people's heads instead. 
 


That's what AI changes. The rules and the exceptions and the edge cases can all be handled now. You can build something that actually reflects how an expert thinks, not a simplified version of it. But someone still has to get there. Asking someone with a full client portfolio to also redesign their profession's working methods, alone, in the evenings, is not a fair ask. Sitting with them for a session and building the first version together changes everything. The second workflow, they build with you watching. The third, they build alone." 

What the work looks like day to day? Walk us through an actual co-build. 

"You start with the work, never with the technology. I ask: what did you do last week that felt like it shouldn't take that long? Then we pick one thing and pull it apart. What comes in, what goes out, what sources do you check, where does your judgment actually enter, and where are you just moving information around. 

Then we build, together. The first version is never right. You run it against a real client, you look at the output, and you iterate the instructions four or five times before it holds. In the process something interesting happens: the accountant sees their own expertise written down for the first time. That moment, when someone says 'that's exactly how I would have done it, and now everyone in my firm can do it like this in no time' is when the scepticism breaks, and for me a really fun part of doing this." 

And the humans stay in the loop? 

"Always. Nothing reaches a client without an adviser's sign-off, and anything touching tax law or regulation has to be verified against primary sources, the actual legislation, not a model's recollection of it. Part of the engineering in accountancy AI engineer is designing exactly where human review sits in every workflow." 

The role also includes training in-house people at the client. Can you explain in more detail? 

"The goal is never to make a firm dependent on us. In every firm you look for the people who were already experimenting on their own, and you develop them into the firm's internal AI architects, the people who build and maintain workflows long after we've moved on. The end state is a firm that improves itself.

My favourite moment is arriving at an accountancy firm and finding workflows I've never seen before, built by someone we trained." 

Who becomes an accountancy AI engineer? What does the background look like? 

"A few years into practice, so the domain knowledge is real. Also, you have to be slightly frustrated. Frustrated in a productive way: you look at how the work gets done and think, ‘this cannot be how we keep doing this, it’s insane!’. Usually you will already have started building things, an automation here or there, a GPT your whole firm is now using... 

What matters is bias toward action. This is a job for people who build things, not people who talk about building things. And you need genuine warmth for the profession. You're asking people to change how they've worked for twenty years. You can't do that from a place of contempt for the old way. The old way made sense once." 

Is this a step out of accountancy? 

"I'd say it's a step deeper in. You stop working on one client portfolio and start working on how the work itself gets done, across many firms. Every pattern you spot in the field feeds back into the product: the workflow library, the roadmap, the platform. You become a primary source of truth for what firms actually need. Very few roles in this profession let you influence it at that scale." 

Legal engineers went from novelty to standard hire in about three years. Same trajectory here? 

"I think so, and possibly faster, because the economics are so direct.

My prediction is that within a few years, mid-sized firms will employ their own AI architects the way they employ IT managers today, and the people doing this job now will have trained them. It's early enough that the people who take these roles this year get to define what the title means. That doesn't happen often in a profession this old." 

Ravical is the agentic operating layer for full-service accounting and tax firms. The billable hour has a ceiling. And as AI makes people more efficient, there are fewer hours to bill. Ravical breaks through it. The platform identifies revenue opportunities across an entire client book, plans and executes the work, and prices it by outcome. Firms grow revenue from existing clients without adding headcount.

We're hiring for Accountancy AI Engineers in London and in Ghent:

Apply for the UK-based Accountancy AI Engineer
Apply for the Belgium-based Accountancy AI Engineer