Product Manager-Enterprise AI Agents

The Role

You will partner directly with the founder to build AI agents that automate high-value ERP workflows finance, procurement, vendor reconciliation, community operations. In a typical week you will:

• Scope a customer workflow and decide if it is agent-shaped (autonomous) or copilot-shaped (human-in-the-loop).

• Design the prompt/tool/action graph for an agent — what tools it calls, what guardrails it has, when it escalates to a human.

• Prototype in Cursor, v0, Bolt, or Lovable — you ship v1 with one engineer, not a 12-person squad.

• Build the evaluation harness: define pass/fail criteria, write LLM-as-judge prompts, run error analysis on production traces.

• Iterate weekly based on real customer usage data, not hypothetical roadmaps.

• Own the agent reliability metrics: accuracy, latency, hallucination rate, fallback rate, cost per task.

• Educate the sales and CS teams on what the agents can and cannot do — set expectations, not hype.

You Are a Strong Fit If

• AI shipping experience: You have shipped at least one production LLM-powered feature or agent, and you can name the architecture, the eval metrics, and the business impact.

• Vibe-coding fluency: You have used Cursor, v0, Bolt, Lovable, or Replit to prototype something real in the last 60 days.

• Agent architecture: You understand tool-calling, function schemas, MCP, retrieval-augmented generation, and agent orchestration frameworks (LangGraph, CrewAI, OpenAI Agents SDK).

• Eval-native thinking: You can design an eval pipeline from scratch — define rubrics, build LLM-as-judge prompts, set up regression tests, and instrument observability (Braintrust, LangSmith, Arize, Langfuse).

• Technical depth: You can read and write Python or JavaScript comfortably. You are not afraid of a terminal, a Jupyter notebook, or a production log.

• Experience: 3–7 years of product management experience, at least 1 year with AI/ML products.

• Location: You are in Hyderabad or ready to be here Monday–Friday.

Strong Pluses

• ERP or enterprise software domain experience (you understand GL entries, procurement workflows, vendor reconciliation).

• Experience at an AI-first startup: Sarvam, Krutrim, Atlan, Observe.AI, Uniphore, Gnani, Yellow.ai, or similar.

• A public artifact: GitHub repo, blog post, Loom, or demo of something you built with LLMs.

• Founding-PM or 0-to-1 product experience.

• Experience with MCP (Model Context Protocol) for tool integration.

What We Will Not Hire

• PMs whose “AI experience” is writing PRDs that mention AI. If your answer is “we integrated ChatGPT,” this is not the role.

• PMs who delegate evals to engineering and never look at production traces themselves.

• PMs who need three weeks to ship a prototype.

• PMs who cannot explain the difference between an agent and a chatbot.