Referment is working with a London-based company that has built an AI platform turning the data behind major construction projects into decisions their leaders can act on — forecasting delay and cost risk on large infrastructure and helping the people running those projects decide what to do about it. The underlying machine-learning capability is trained on one of the largest datasets of construction schedules in existence.

The company runs product differently from most. AI has collapsed the cost of building software, so its teams are organised into small squads of two to three people, and everyone in a squad is a contributor — there’s no “write the spec and wait”. As a Product Manager you’ll lead one or more squads, own a quarter’s mission, reconcile priorities when unplanned work lands, and make real trade-offs.

And you build. You’ll prototype with AI tools, direct agents, review and edit generated code, and take ideas to working tests in days. Increasingly the product itself is agents, so you’ll design what they do, what they must never do, and where humans stay in the loop.

The Role

• Own your squads’ quarter: plan it, reconcile it against unplanned work, and deliver it, making build/kill decisions on evidence.
• Build rather than brief — write clear specs, prompt agents, and review and edit generated code and output.
• Design and ship AI agents, defining their behaviour, guardrails and the context and tools they use.
• Run structured evaluations — define what good looks like before building, then prove it with representative datasets and scoring rubrics.
• Help people decide under uncertainty, turning risk and forecasts into something a busy executive can act on.
• Get close to customers through user and buyer research and client meetings.
• Lead contributors rather than a process, coordinating dependencies across squads.

What We’re Looking For

• Two-plus years owning a B2B or enterprise AI or SaaS product — accountable for a product area, with shipped features real clients depend on.
• You already build with AI daily: prototypes, analysis, and working code you review and edit rather than write from scratch.
• You’ve built with LLMs in earnest — enough to reason about model choice, prompting, context, retrieval and tool use.
• You can take an ambiguous problem, get to what customers actually need, and make the right calls through understanding the trade-offs.
• You’ve killed things — you can point to experiments you ran and the roadmap items that died as a result.
• Fluent with data, including designing structured evaluations rather than only reading dashboards.
• Sharp communication: able to make a complex trade-off legible to an engineer, a customer and a leadership team member.

Nice to Have

• Experience with a product built on data science or machine learning, and comfort discussing forecasting and uncertainty.
• Experience shipping to senior operational or finance buyers.
• Some understanding of construction, planning and/or risk.
• A formal technical background.

This could suit a product manager from an AI or data-heavy B2B product, or someone with a strong technical and analytical background who wants real ownership in a small squad.

Referment