Product Manager, AI Agents & MCP Tools

Recorded Future•Boston, MA

2d•Remote

About The Position

With 1,000+ intelligence professionals serving over 1,900 clients worldwide, Recorded Future is the world’s most advanced, and largest, intelligence company! We're looking for a hands-on Product Manager to own the lifecycle of Recorded Future's AI agents and MCP tools — the intelligent workflows and tool integrations that apply large language models to real threat intelligence problems for our customers. This is not a role focused on training models or building LLMs. Instead, you'll build, evaluate, and ship agents that orchestrate existing models, and shape the MCP tools those agents and our customers rely on to access Recorded Future intelligence. You'll be responsible for the full arc of an agent — from initial build, through evaluation and iteration, to customer testing and deployment — as well as the quality and coverage of our MCP tool surface. You'll write and maintain evals, refine tool descriptions, identify gaps in tool coverage, analyze how tools are actually used, and make the call on what ships. This role suits someone who is technically fluent, comfortable getting their hands dirty in GitHub and prompt engineering, and grounded in strong product judgment about what's worth building. This isa mid-level role for a builder who moves fast, tests rigorously, and cares more about whether an agent or tool solves the customer's problem than whether it demos well.

Requirements

Agent Builder with hands-on experience building LLM-powered agents or workflows — or the technical aptitude to ramp up quickly.
Model-Literate , with working knowledge of major LLMs and a practical sense of their pros, cons, and cost tradeoffs.
Tool-Design Sense , able to write clear tool descriptions, reason about how agents select and invoke tools, and spot coverage gaps.
Technically Comfortable , able to work in GitHub, read and reason about code, and engage credibly with engineers on agent and tool design and evaluation.
Evaluation-Minded , understanding how to define quality, write evals, and use them to drive iteration rather than relying on vibes.
Data-Informed , comfortable analyzing usage patterns to guide decisions about what to refine, build, or retire.
Customer-Oriented , skilled at working directly with users to validate that what's built actually solves their problem.
Strong PM or Business Analyst Skills , able to prioritize, define requirements, and connect technical work to business outcomes.
Pragmatic and Outcome-Driven , comfortable shipping, measuring, and improving in fast iteration cycles.
3–5 years in product management , technical program management, or a hands-on technical role building AI/LLM-powered products, agents, or tool integrations

Nice To Haves

Familiarity with MCP (Model Context Protocol) or similar tool-integration frameworks is a plus.
Cybersecurity experience is a plus but not required — provided you can ramp up quickly on the domain.

Responsibilities

Own the end-to-end lifecycle of AI agents — design, build, evaluation, customer validation, and deployment.
Build agents that orchestrate LLMs and tools against real intelligence use cases, selecting the right model for each task based on capability, latency, and cost tradeoffs.
Own and refine the MCP tool surface — writing clear, effective tool descriptions, identifying gaps in coverage, and improving how tools expose Recorded Future intelligence to agents and customers.
Analyze MCP tool usage patterns to understand what customers and agents actually invoke, where tools fail or underperform, and where new tools are needed.
Write, maintain, and expand evaluation suites to measure agent and tool quality, catch regressions, and guide iteration; update agents and tools as models, data, and customer needs evolve.
Test agents and tools directly with customers, gathering feedback and confirming that outputs meet their workflows and expectations before and after launch.
Work hands-on in the codebase (GitHub) alongside engineers — reviewing changes, prototyping, and contributing to agent logic and tool definitions where appropriate.
Maintain a working understanding of the LLM landscape, tracking the strengths, weaknesses, and cost profiles of available models to make informed build decisions.
Define and track agent and tool performance metrics — accuracy, task completion, tool invocation success, latency, cost per task, and customer satisfaction.
Prioritize the agent and tool roadmap, focusing effort on the capabilities that deliver the most customer value.
Partner with intelligence, engineering, and design teams to ensure agents and tools integrate cleanly into the broader platform and customer experience.
Establish repeatable practices for building, evaluating, and shipping agents and tools reliably and safely.