Staff Product Manager (Artificial Intelligence)

• The Product team is at the forefront of Brex’s mission to empower employees anywhere to make better financial decisions. With a deep understanding of the business, we identify and scope out the most impactful opportunities for Brex to tackle

• We are responsible for aligning cross-functional teams — such as Engineering, Legal, Compliance, and Design — on key decisions. We set strategy and drive products from inception to launch, enabling Brex to grow rapidly and help our customers reach their full potential

• You’ll work on our core AI team building agents for the finance admin and own one of these end to end. You’ll set its product direction, decide which capabilities to build and how they should behave, and spend real time with controllers and finance teams to learn what they’ll actually trust an agent to do on their behalf

• From there you go as deep as it takes to make it real: shaping agent behavior, writing and iterating prompts, designing the evals that prove it works, and digging into traces when it doesn’t. You’ll partner closely with engineering, design, and research, and you’ll set the quality bar for agents that customers rely on to keep their spend clean and their numbers right

• Own product direction for your area of the Finance Admin AI surface — the strategy, the roadmap, and the day-to-day calls on what ships

• Translate messy, company-specific finance workflows — expense policies, review and approval chains, month-end close, the way each customer actually enforces its rules — into agent behavior that holds up in production

• Define what “good judgment” means for an agent operating on someone else’s money, and build the evals, prompting approaches, and feedback loops that get it there — then re-invent them as the underlying models and tooling shift

• Get hands-on. Prototype, measure, read transcripts and traces, and iterate toward what actually works rather than what looks good in a demo

• Earn autonomy incrementally: figure out where the agent should recommend, where it should act, and how a customer builds enough trust to let it move from one to the other

• Work directly with finance admins at real companies, and let their usage — not assumptions — drive the next iteration

• Set and defend a high quality bar for agents operating in a trust-critical, financial context

Benefits

• Medical, dental & vision

• Generous vacation

• Spring Health & Calm

• Year-end company shutdown

• Milk Stork for nursing parents

• In-office days, lunches & offsites

• Commuter benefits

• 4 fully remote weeks

• Carrot Fertility benefits

• Generous parental leave- Depth in company workflows or in data and reporting, and genuine interest in the other — the queues, approvals, and exceptions an admin works through, or how people get answers out of their own data and what it takes to model it so those answers are right

• You’re an experienced product manager. You have a track record of owning complex, user-facing products end to end — setting direction and shipping to real users — with the scope and judgment this senior role calls for

• You’re comfortable with ambiguity and with systems you can’t fully predict. You pull signal out of noise, make good intuitive bets, and ship and refine rather than holding out for a perfect answer. Perfect is the enemy of good here

• You move fast and go deep — comfortable prototyping, reading data, and getting into the details rather than working through layers

• Strong product taste and judgment about what’s worth shipping to customers and what isn’t

• You’ve built for company admins, and you’ve built it for customers. You’ve shipped software to the personas who keep a company running — finance, accounting, HR, payroll, IT, compliance, data — and you know how much this can vary from one company to the next. It matters that this was product you sold to other companies, not an internal system inside your own

• You have technical depth, and you’ve worked on systems without a single correct answer. Having built agentic software is a strong plus. Probabilistic, optimization-driven systems where “done” doesn’t exist and the work is making it steadily better: AI/ML, search and ranking, marketplace matching, risk and fraud detection, game engines. You’re comfortable when the spec is an outcome and a set of signals rather than a decision tree. Also, if you’ve shipped an agent that real users depended on — designed the evals, iterated the prompts, worked out where it should act on its own versus stop and ask, and lived through the failure modes — say so and tell us what you learned

• Hands-on depth in prompting or eval design; familiarity with fintech, expense management, or accounting workflows