Do you need an AI governance hire before Series B?
- The failure mode is a missing owner, not a missing hire.
- An early owner has authority and discipline; compliance background is optional.
- The role cannot rotate — linkage between answers and evidence dies with turnover.
- Hire when two or more signals compound: weekly questionnaires, regulated pipeline, EU high-risk exposure, live incident, or contractual response SLAs.
- Hire a pragmatic operator who produces evidence as a by-product; resist the optics hire.
What actually needs owning
Strip the title away and the work is concrete: keep the inventory of AI systems true, keep each system's classification and evidence current, own questionnaire responses end to end, watch the two or three regulatory frameworks that touch your market, and run the incident path if something goes wrong. At ten or thirty people, that is hours per week, not a job. The failure mode is not the missing hire; it is the missing owner, with the work scattered across whoever seemed least busy when each questionnaire arrived.
Who holds it early
The natural early owner is a founder, COO, or operations lead with security-adjacent instincts, backed by counsel on retainer for the genuinely legal questions. The owner does not need a compliance background; they need the authority to get answers out of engineering and the discipline to record decisions when they happen. What the role cannot be is rotating: an owner who changes quarterly takes the linkage between answers and evidence with them each time, and the record quietly dies.
The signals it is time to hire
Hire when the work outgrows a fraction of a week, and the signals are unambiguous: questionnaires arriving weekly rather than quarterly; regulated-sector buyers (finance, health, public sector) becoming the pipeline rather than the exception; an EU footprint with systems likely to classify as high-risk; a live incident that exposed the process as one person's memory; or enterprise contracts that now impose governance obligations with response-time commitments you must staff to honour. Two or more of these and the fraction has become a role.
What to hire when you do
Hire a pragmatic operator with compliance literacy, not a policy author. The first governance hire's job is to build systems that produce evidence as a by-product of normal work: inventories that update, answers wired to their sources, reviews that happen on a calendar. A hire who measures output in documents produced will bury you in binders that reviewers do not believe. And resist the optics hire: a "Head of AI Ethics" appointed before the substance exists is theatre, and theatre is discovered in review.
Key terms
- Accountable owner
- The single named person who holds AI governance work end to end, whether or not it is their whole job.
- Review volume
- The rate at which enterprise security and AI questionnaires arrive; when weekly, the work outgrows a fraction of a week.
- Regulated pipeline
- Finance, health and public-sector buyers whose diligence depth turns governance from occasional work into a role.
- Optics hire
- A senior appointment made before the underlying substance exists; discovered in review and read as theatre.
- Evidence by-product
- Governance built so evidence is generated in the course of operating, not produced in a scramble before an audit.