Staff Inbound Product Manager, AI Specialists and Voice AI

About the role You will own a piece of that portfolio outright. What it does, who it is for, what “good” means, and whether we are honestly there yet. The interesting part of this job is not the model. It is everything around the model. Every customer environment is different: different knowledge quality, different catalog, different naming conventions, different appetite for letting an agent act on its own. An agent that performs great on a clean eval set can behave differently in a live enterprise for reasons that have nothing to do with reasoning ability. Understanding why, and designing for it, is the job. We treat evaluation as a product surface rather than a QA step. We run production-ready code against real customer data before we ship, we write and tune our own judge prompts, and eval results are our launch criteria. ServiceNow moves faster than a company our size has any right to. We shipped an autonomous L1 specialist, and the evaluation discipline behind it, in the time most enterprises spend forming a committee. We are not reacting to AI. We are deciding what AI-first, enterprise-grade software actually means, for the largest companies in the world, who run their business on our platform and cannot afford for us to get this wrong. A lot of that definition does not exist yet, which means on many weeks you will be the person framing the problem rather than receiving it. How we work is changing as fast as what we build. We are AI-native in our own process, not just in what we ship: prototype with AI instead of describing the idea, put a working demo in front of people instead of a slide about a demo, generate the deck, draft the spec, stand up a rough eval harness. ServiceNow invests in state-of-the-art tooling and expects you to use it, so the role itself is being redefined in real time, and product managers are better positioned than anyone to lead that shift and pull an AI-product team forward with them. We prove products out in the wild before we scale them. That means working forward deployed: embedded with a customer, in their environment, building and tuning against their real data and their real edge cases until the it works. Then the harder half, turning what you learned into something that generalizes to the next thousand customers. At most AI companies those are two different jobs, and the handoff between them is where the insight gets lost. Here it is one person, and it is you. What you get to do in this role Own an AI-Native product end to end, from the customer problem through launch criteria to how it performs in production. Deploy forward. Embed with some of the largest IT organizations in the world on problems they have not solved, build in their environment against their real data, then get into the guts of the AI Agent execution logs to understand what is actually happening. Both halves matter. The best insights live in the gap between them. Define what “good” means for an autonomous agent, alongside engineering and our AI Quality CoE, then hold the line on it. That includes the hard conversations about what we can defensibly promise a customer and how much value we are actually delivering them. Own your outcome metrics and make sure the instrumentation exists to know whether what you shipped moved them. Run design partnerships with enterprises betting real service desks on this, and get what you learn back into the roadmap fast. Write the strategy, the specs, and the analysis yourself. Work AI-native. Use AI tools to prototype, draft, and communicate. Build a working prototype instead of describing one, and set the bar for how the team around you works. Coach other product managers. This is a senior role on a small team, so how you raise the people around you counts as much as what you ship. To be successful in this role you have: Have owned an AI-Native product end to end and been the person accountable for its quality bar. For most people this means eight or more years in product management, but we care about the ownership, not the years. Have shipped something built on LLMs to paying customers, and understand what enterprise-grade means for AI: the quality bar, the guardrails, the telemetry, and the work of earning a customer’s trust before they hand an agent real work. Can do your own analysis. You should be comfortable pulling your own data and understanding how a metric was built before you rely on it. Stay curious, and can find the signal in the noise. Document your position clearly and justify your approach. Decisions here get made through research and documentation. Are willing to own the outcome when the org chart is ambiguous about who should, which happens often. Preferred qualifications Agent evaluation: eval set design, LLM-as-judge, closing the gap between offline results and production behavior. Forward-deployed, applied AI, or embedded delivery experience. You have built something inside a customer’s environment and then had to make it work for everyone else. IT service management, service desk operations, or enterprise support. Voice agents, or conversational products where latency and turn-taking matter. Experience taking a product from zero to one, where the users, the quality bar, and the economics all had to be worked out at once. AI-native working habits. AI assistants and productivity tools are already part of how you work. You use them to prototype, draft, and analyze, and you have automated enough of your own day-to-day that you spend your time on the judgment calls instead of the mechanics. Bonus if you have changed how a team around you works because of it.   FD21       For positions in this location, we offer a base pay of $166,500-$291,400, plus equity (when applicable), variable/incentive compensation and benefits. Sales positions generally offer a competitive On Target Earnings (OTE) incentive compensation structure. Please note that the base pay shown is a guideline, and individual total compensation will v