Jun 18, 2026

11 min read

Blog

How We Built Adherent

This blog was originally posted on 18th June, 2026. Further developments may have occurred after publication. To keep up-to-date with the latest compliance news, sign up to our newsletter.

AUTHORED BY SIOBHAN FAIRMAN, VP OF ENGINEERING, ADHERENT

The companies we serve make the products in your kitchen, your car, your child’s bedroom. They sell into 195 countries.

Each one is trying to do the same hard thing: comply with a wall of regulation that grows faster than any team can keep up with – so the products that reach you, your family, and the planet are safer, cleaner, better.

That has always been the mission.

Adherent is not a rebrand. Adherent is what 18 months of rebuilding looks like when a company decides to deliver on its mission a different way – faster for the people doing the work, smarter, validated by twenty years of human expertise – and earns a new name on the other side.

Here’s the honest version of how we got here.

November 2022

When ChatGPT landed in November 2022, a small group of us inside the company saw it for what it was.

Compliance has one problem at its core that has defeated every wave of technology that came before it: applicability. Which rules actually apply to this product, in this market, for this company? Hand-curated databases, rule engines, classical NLP, classical machine learning – we’d worked with all of them. Each took us a long way. None of them quite cracked the full problem.

LLMs were different. Within weeks we could feel it. So we did the only sensible thing: we stopped speculating and started experimenting.

And to do any of it, we needed a team that didn’t exist yet. Building an AI team from scratch in 2023 meant learning, in real time, what we didn’t even know we didn’t know. What skills actually mattered. Which roles we needed first. Where the lines were between research, engineering, and product. How to spot the gap between someone who’d put “AI” on their CV and someone who could actually move the needle on our problem. We mis-stepped, corrected, mis-stepped again. Some of the people we brought in were industry veterans who anchored us. Some were younger engineers who showed us what was actually possible when you let go of how things had always been done. Some surprised us with skills we hadn’t been hiring for and didn’t know to ask about. The mix we have today looks nothing like the team we’d have drawn on a whiteboard at the start. That’s the point.

For the next 18 months, a quiet, disciplined pilot programme ran inside the company. Regulatory extractors built side by side with our SMEs. Requirement extraction prototypes benchmarked against the approaches we’d been trying for years. Translation engines tested against our human translators. OCR pipelines tested against documents we already knew the answers to. Every experiment had a control. Every result had an expert in the room.

We weren’t trying to ship a product. We were trying to answer one question: was this the thing we’d been waiting for, or was this another almost-thing?

By the summer of 2024 we had our answer. It was the thing. Not in a slide, not in a hype cycle – in our hands, on our data, validated by the people who had spent their careers getting these answers right.

The Diagnosis

In those same weeks, one of our engineering leads stood up at a Product-Engineering offsite and asked the right question, out loud, in front of the people who could decide: if we were going to build something fundamentally new – agentic AI, designed from first principles for what compliance is becoming – where would we build it? Inside the institution we’d spent twenty years building, or alongside it?

He didn’t pretend either answer was easy. He laid out the real economics of building agentic systems on a platform designed for a pre-AI world, and the real economics of starting fresh. Then he closed by saying: pick.

The Room in San Francisco

In September 2024, eight of us flew to San Francisco for three days with our investors at Luminate Capital. We had one question on the table: keep investing in the platform that had made Compliance & Risks the category leader for two decades – the platform our customers count on every day – or also build something new, from a blank page, alongside it.

We walked into that room knowing what the answer was, and knowing how hard it would be to say out loud. By the time we left, we had agreed: keep our flagship strong for the customers who rely on it, and at the same time build a new new agentic AI product alongside it – starting with sustainability, because solving applicability and requirements in sustainability would give us the platform we’d need for the much harder problem of Product Compliance.

We gave it a codename. Skywalker. We named the bets we’d make and the bets we’d defer. Our CEO drafted the company-wide email we’d send the following week, with the line: “We are super excited to share how we will accelerate the realization of our Big Hairy Audacious Goal.” We didn’t know yet whether the company would believe us.

Outside the Room

That same week, on the other side of the city, Salesforce opened Dreamforce by welcoming the world to Agentforce. The biggest SaaS company on earth was telling its customers, from the stage: this is real, this is where we all have to go.

The decision we’d just made inside a room of eight people was the same decision Salesforce was announcing to a stadium 1 mile away. We took some quiet confidence from the man on that stage.

The Bet

We chose sustainability as our first product not because it was easy, but because it was the right place to lay the foundation – for our customers and for us. Every customer we work with has a sustainability problem that gets bigger every year: more reporting frameworks, more jurisdictions, more scrutiny, more stakeholders. They are trying to make products the planet can live with. We could help them do that faster.

Sustainability and Product Compliance share the same hard problem at the bottom: applicability – which rules actually apply to me? Solve that for sustainability, build the platform underneath it, ship it to real customers, learn from them, and you’ve earned the right to attack Product Compliance – which is an order of magnitude harder, and just as important to the world.

The first preview environment of the new product went live on Vercel on October 14, 2024. Twenty-six days after we got back from San Francisco.

That speed wasn’t an accident. It was an investment thesis. We told our investors, our board, and ourselves that we were going to spend less time arguing about architecture and more time putting capability into our customers’ hands. By January 2025 we were running automated security tests on the new app every single night. By mid-2025 we had paying customers on the new product. We did a soft launch over the summer, and we hard-launched the product to the world in October 2025 – and in the 90 days that followed we quickly started rolling out to new customers who are now meeting their sustainability obligations faster, with more confidence, and with far less of their team’s time burned on the mechanics.

The People Transformation

This is the part of the story I find most beautiful, and the part that gets undersold every time we try to tell it.

We didn’t bring in a new team to build the new product. We upskilled the team we had.

Engineers who’d built their careers on mature enterprise stacks were picking up Next.js, Nx, Vercel, Clerk, Postgres and shipping. The reason they could move that fast is the second part of the transformation: we stopped pretending AI was a side project and started treating it as the way we learn.

When Claude Opus dropped in late 2025, something shifted that I don’t think we had words for yet. Over the Christmas break, without anyone asking, our team built things on their own time. A new contractor management app from our CTO. A new front end from one of our Engineering Managers. A soccer team performance tracker now used by three local clubs. A personal product shipped to a paying audience. I built a game-theory app for my kids.

A few weeks later, one of our senior engineers sent me a note: “I implemented a copy requirement yesterday in one day. Otherwise it would take a few days. I feel very powerful with Claude – super powers!”

Our Chief Architect has a phrase I keep stealing: “Go slow to go fast.” You can prompt your way to a feature, or you can spend the time defining the skills, the plans, the engineering practices, and then prompt. The team that has done the second is shipping at a rate I genuinely could not have imagined 18 months ago. And every day faster is a day our customers get to act sooner.

I told this team they weren’t going to become dinosaurs on my watch. This is what it looks like.

Permission to be Curious

If you want a workforce to use AI, you have to make it safe to use AI.

We could have written an AI usage policy that read like a wall. Instead we wrote one that read like a door. Less “you may not,” more “here’s how, here’s what to watch for, here’s how to bring us in early if you’re not sure.” It treated curiosity as the asset, not the risk.

What happened next surprised even us.

At our 2026 Company All Hands, we filled the stage with our innovators. Sales using AI to research prospects faster than anyone in our market. Regulatory specialists extracting impossibly long chemical names from dense source documents in seconds. Marketing, finance, operations – every function had a story. Ideas about how to move faster – for ourselves and for our customers – were no longer trickling up from a few enthusiasts. They were arriving from everywhere, all at once.

A policy that opens doors instead of building walls is one of the best decisions we made all year.

Our Experts

The other half of our edge isn’t engineering. It’s our compliance experts.

Twenty years of human expertise – the deepest bench of compliance and sustainability subject-matter experts in this industry – sits inside Adherent in a way that nobody else can claim. The temptation, with AI, is to treat that as legacy. We did the opposite. We paired our experts with our engineers and invested company time codifying their thinking – capturing how they reason about regulations, what they look at first, what they trust and don’t.

And then we kept them in the loop as our benchmarking and validation experts. Their job didn’t shrink. It deepened. Because in a world where every AI company claims accuracy, ours is the one validated against the people who actually wrote the answer key. Which means the customer who signs up to Adherent is getting answers built by the people who have spent their careers getting compliance right.

Why Adherent?

So, that’s what we’re announcing today.

Adherent is the name we chose for what we’ve become – a company that builds agentic AI compliance products on a modern platform, validated by the deepest bench of compliance expertise in this industry, shipping at a velocity our previous selves wouldn’t have believed. Built so the people who comply for a living can do it faster, with more confidence, and at the scale the modern world demands of them.

The work matters. Every regulation our customers meet, every market they open responsibly, every product they ship that’s a little safer, a little cleaner, a little better – that’s the world we’re trying to help them build. That has always been the mission.

The name isn’t the thing. The thing is the team who chose to relearn their craft, the experts who chose to share what they know, and the customers who chose us before any of it was proven.

It blows my mind what we’re doing together. And wait till you see what’s to come.

Welcome to Adherent.