Where AI belongs in a regulated pharma workflow — and where it doesn’t
The judgment call that decides whether a pilot survives audit. Who owns it, and how they’re hired.
Draft. Figures marked like this are illustrative and pending verification against Recruise placement data & Sachith sign-off before publication.
Key takeaways
- In a regulated pharma workflow, the question isn’t whether AI can do the step — it’s whether the step survives an audit once AI has touched it.
- Every viable pilot has one person who owns the line between assistance and accountability — and that ownership has to be a named, senior seat.
- You hire for it by finding someone who has carried regulatory consequence before, not someone who is merely fluent in the tool.
The audit, not the capability, decides where AI belongs.
In a pharma GCC, the interesting question about AI is almost never “can it do this.” It usually can. The question is whether the resulting step can be defended when an auditor asks who was accountable, what was checked, and how you’d know if the model was wrong. A workflow that can’t answer those questions doesn’t fail the pilot — it fails inspection, later, when the cost is far higher.
So the boundary between where AI belongs and where it doesn’t isn’t drawn by technical feasibility. It’s drawn by traceability and accountability. Assistance in drafting, triage, or surfacing anomalies tends to survive scrutiny. Anything that becomes the unreviewed basis of a regulated decision tends not to — and the difference between those two is a judgment call someone has to own.
“A pharma pilot doesn’t die in the demo. It dies in the audit — and the person who could have prevented that was the hire the centre didn’t think it needed.”
Sachith Rai · MD & Founder, Recruise
Someone has to own the boundary — and it’s a specific hire.
The pilots that survive have a named owner for that boundary: a person senior enough to decide which steps AI may touch, which it may only assist, and which it stays out of entirely — and credible enough that quality and regulatory functions accept the call. This is not the data scientist who built the model and it is not the QA generalist. It’s a distinct seat that sits between them.
What makes the role hard to fill is that it demands two things that rarely coexist. The person needs enough fluency to understand what the model is actually doing, and enough scar tissue from regulated environments to know where consequence lives. Find both in one candidate and you can scale the pilot; find only the first and you have a fast pilot that won’t survive contact with an inspector.
Hire for consequence carried, not tools mastered.
When we search for this seat, the strongest signal isn’t AI experience — it’s a history of owning decisions that had to hold up under regulatory scrutiny. Someone who has stood behind a call in front of an auditor understands the boundary in a way no tool training conveys. The AI fluency can be built on top of that; the instinct for consequence cannot be built in a quarter.
The centres that get this right treat the boundary-owner as a foundational hire, made before scale rather than after a scare. The ones that get it wrong scale the tool first and go looking for the owner only once an audit exposes the gap — by which point the cost of the missing seat has already been paid.
One pattern worth knowing, every week.
The Signal is our weekly read on the senior GCC talent market — one chart, one pattern, no noise. Written from live placement data.
More from Recruise Insights.
Most GCC AI pilots stall at the same place. Here's why.
We've mapped GenAI rollouts across the GCCs we work with. The failure mode is consistent — and it isn't the technology.
GenAI hires aren't the same as AI/ML hires. Most JDs miss it.
Three distinct profiles are being confused for one. The mis-hire rate is predictable. The fix is upstream of recruitment.
The roles AI is quietly creating inside GCCs
Not the ones in the headlines. The senior seats that appear when a function moves from doing the work to governing it.