Senior Product Manager, AI Platform

Decide what deserves to exist Aerogram sits between a core product and the reality of implementing it inside complex enterprises. Each side shapes the other. What we learn from one deployment can expose a missing capability, challenge an assumption, or reveal a pattern that matters across many customers. Your job is to turn those signals into product direction. You'll work across customers, FDEs, engineering, and leadership to decide what becomes part of Aerogram, what stays specific to one implementation, and where what we learn should change the product itself. Those calls won't always be right the first time. What matters is knowing what you expected, watching what actually happens, and adjusting quickly. Over time, that judgment is what keeps Aerogram coherent while making it increasingly useful to the enterprises we serve. What you'll do Make the call between product and bespoke. Every week something we built for one customer is a candidate to become platform. You decide, you write down why, and six months later we'll open that document together and see if you were right. Get signal nobody hands you. Our users can't show you their data, their screens, or sometimes their office. You'll get the truth anyway — through FDEs, redacted samples, and sitting where the work happens. And you'll learn to hear the difference between the buyer, the user, and the person who can quietly kill the whole thing; they are never the same person. Prototype instead of specifying. A written spec is a terrible way to explain an agentic workflow to anyone. Turn the idea into something a user can react to in days, yourself, with whatever tool gets you there. Production code belongs to engineering. Decide what "good enough" means — and put a number on it. An eval set, a target, the failure modes we accept and the ones we never will, where the human checkpoint sits, what the audit trail shows. This gets written before launch, and the customer's risk team will read it more carefully than anything else we send. Make it pay for itself. Cost per workflow run, margin per account once implementation effort is counted, where platform scope ends and billable work begins. Kill the features that can't carry their own weight — before sales promises them, not after. Judge from small numbers. There is no A/B test coming to save you. Five conversations, one usage pattern, and your judgment — then a consequential call, and the answer to whether you were right arrives months later. If that sounds terrifying rather than interesting, stop reading here. What we're looking for You've shipped a model-driven feature to real users and measured what happened. One is enough. One in production beats three in demos. You've killed something you believed in — and can name the evidence that changed your mind, and when it arrived. "Leadership deprioritized it" is not that story. You prototype. Show us something you built yourself to settle an argument. Technical enough to hold the room — model tradeoffs, latency budgets, what a deployment constraint does to a design. You can follow a pull request. You don't need to write one. Senior people listen to you. Frontline users are honest with you. You notice when the two want opposite things. You think in margins without being asked. What it costs to serve, what it costs to generalize, who pays. Helpful, not required: enterprise software, regulated industries, or having watched a security review eat a roadmap. Don't apply if you need the roadmap handed to you, or you measure your work by what ships rather than what holds up in month ten. About Cleverse We're a venture builder: we see where the digital world is going, build the business model around that vision, and craft the product that makes it real. Our teams have shipped the kind of work the industry remembers — built when most thought it couldn't be done at that scale. We host the rooms where the country's builders gather, and we tend to be standing where the field is about to go, not following it. Aerogram is our current bet — on the gap between AI that demos and AI that does real work inside a business. We stay true to the outcome a customer is paying for, not to any one way of reaching it, and we're never quite satisfied: there's always a version of the work that's better than the one we shipped. About Aerogram Aerogram is an AI workflow platform for organizations where mistakes have consequences — banks, asset managers, insurers, government. It takes the work that runs a business — documents moving between systems, reports assembled by hand, approvals crossing five desks — and puts AI inside it, with the accuracy, audit trails, and human checkpoints those organizations answer to regulators for. It runs where the customer needs it to: our cloud, theirs, or entirely inside their own walls where we never touch production. We deploy through Forward Deployed Engineers who build the first version on-site with the customer's own team — which is why we know what enterprises actually do all day, not what they say in requirement documents. The workflows Aerogram runs are the ones a customer escalates about when they break. That's the bar, and it's the whole point.