The client
Series B enterprise SaaS in London.
Around 200 people. Workflow automation for financial services.
Revenue growing well. Solid engineering team. Real customers.
Board pushing for “an AI strategy.”
Why the client came to us
The internal talent team was strong at hiring software engineers. But AI leadership was a different game entirely.
They'd already talked to two candidates through their network. Both were impressive presenters. Neither had shipped anything to production.
The CEO realized he needed someone who could tell the difference between an AI strategist and an AI builder. That's not a recruiter skill — that's a technical evaluation skill. That's us.
The task
Find a Head of AI to build AI capabilities into the core platform.
The core problem
The brief was vague: “we need someone to lead AI.”
When we dug deeper:
- CEO wanted to "bolt on LLM" to everything
- No clarity on which use cases would actually move the needle
- Solid engineering team but zero ML infrastructure
- No data pipeline for training
- No clarity on what AI should do for the product
People with AI in their headline are everywhere. People who can build an AI function inside a business — much fewer.
Additional complexity
1. Vague strategy
The client wanted “someone to do AI” without defining what AI should do. We challenged this directly: you need 2-3 use cases that impact retention and expansion revenue, not “an AI strategy.” We ran a working session with the CEO and CPO. Two areas surfaced fast: document processing — the ops team was drowning in manual review — and churn prediction, where the product had enough usage data to be useful.
2. Wrong job description
The original JD said “define and lead the AI strategy.” A generic “Head of AI” JD would attract AI strategists, not builders. We killed it and rewrote it: “Build intelligent document processing and predictive churn scoring into the core platform.” That one change filtered out most strategy consultants who would have taken the role, done a roadmap presentation in month one, and left the engineering to someone else.
3. “AI strategist” candidates
The CEO liked three profiles early on. Strong brands on paper. Fluent AI narratives in interviews. We rejected all three. None had shipped a model to production in a similar environment. One spent two years making AI strategy decks at a consulting firm. Another talked about “LLM transformation” but couldn't describe a single pipeline he'd built. The client pushed back. We held the line.
References told the full story. In every case, these candidates had been hired to “lead AI” and had produced a strategy deck, a team request, and a timeline — but no shipped product. We've seen this pattern enough times to recognize it in the first conversation.
What this person actually does
This wasn't an “AI strategist.” This was a builder.
Product AI
Intelligent document processing, NLP pipelines for financial services data.
Predictive
Churn scoring, expansion signals.
Infrastructure
Data pipeline from scratch, model monitoring.
Team
Hire the first ML engineers, set technical direction.
Stakeholder
Translate AI capabilities into product roadmap.
Process
7 weeks
Time to close
22
Profiles shared
3
Rejected by us
Month 3
First feature
References focused on: “Did this person build something that went to production?”
The result
The hire was a former ML lead from a mid-stage B2B SaaS.
Built document extraction and NLP pipelines in production. Not glamorous AI — exactly what the product needed.
Within 4 months:
- intelligent document processing shipped (manual review reduced 60%)
- churn prediction pilot started
- first ML engineer hired
Business impact
The company didn't just get a hire. They got clarity.
- Manual document review reduced by 60% — freeing the ops team for higher-value work
- Churn prediction pilot gave CS team early warning signals for the first time
- AI became a concrete product capability, not a board-deck buzzword
- The hire's first ML engineer joined within 4 months — team building had begun
- Board stopped asking "what's our AI strategy" — they could see it shipping
Our value
- Stopped the client from hiring the wrong kind of leader
- Challenged the vague "do AI" strategy and made it specific
- Killed the generic JD and rewrote it around real use cases
- Rejected candidates the client liked — and were right to
- Found a builder, not a strategist