Product Manager - Conversational AI

Job Description

MyOperator is building AI-driven conversational surfaces — chatbots and voicebots that handle real business conversations, run structured journeys, pull from knowledge bases, and hand off to human agents when needed. We're looking for a Product Manager to own this domain: someone who genuinely understands how LLMs and RAG (Retrieval-Augmented Generation) systems work, and can turn that understanding into product decisions — what to build, which model tradeoffs to make, and how to know if a chatbot or voicebot is actually working.

This is a technical AI product role. You'll be defining journeys, knowledge base behavior, and evaluation criteria for systems where outputs are probabilistic, not deterministic — so the ability to reason about model behavior, failure modes, and retrieval quality matters as much as classic product skills.

About MyOperator

MyOperator is a Business AI Operator platform that allows businesses, teams, and AI Agents to work in tandem for customer operations, i.e., handle Sales, Support, Escalation, Feedback, and Refund processes. With over 12,000+ businesses using our platform, we are the largest in the space.

MyOperator is built for people who want to work on ambitious problems at a meaningful scale. We value ownership, speed, critical thinking, and a bias for building things that create real customer and business outcomes. This is a high-expectation, high-learning environment where people are trusted to think independently, challenge ideas openly, move with urgency, and keep raising the bar as we build for long-term impact.

Key Responsibility Areas

• Own the conversational AI product roadmap — chatbot and voicebot journeys, knowledge base/RAG behavior, and model selection — writing specs that account for real LLM behavior and failure modes, not just ideal-case flows.
  

• Define and maintain how success is measured — partner with engineering to build and evolve eval systems that track quality (accuracy, containment, hallucination rate) so every release has a clear bar to clear.
  

Requirements — Must Have

• Strong practical understanding of how LLMs work — prompting, context windows, tool/function calling, and their real-world limitations in production systems.
  

• Solid understanding of RAG (Retrieval-Augmented Generation) systems — how retrieval, embeddings, and knowledge bases feed into model responses, and where these systems typically break down.
  

• Working knowledge of the AI model landscape — tradeoffs between models on cost, latency, and accuracy — and how to translate that into build/model-selection decisions for a product.
  

• Demonstrated ability to write specs for AI-driven product surfaces (chatbot/voicebot journeys, escalation/handover logic) that explicitly account for probabilistic model behavior and edge cases.
  

• 4+ years as a Product Manager, with meaningful hands-on time owning an AI/LLM-powered product end to end.
  

Requirements — Good to Have

• Hands-on background building or shipping chatbots or voicebots — design, integrations, and deployment.
  

• Experience designing conversational journeys — structured, multi-step dialogue or task flows for chat or voice agents.
  

• Experience building or defining eval systems/frameworks to measure chatbot or voicebot success (e.g., accuracy, containment rate, hallucination rate).
  

• Familiarity with prompt engineering and iterating on prompts/journeys based on eval results.
  

• Experience designing human-handover or escalation flows between a bot and a live agent.
  

This profile is not for

• Candidates who treat "AI feature" as a checkbox without understanding how LLMs and RAG systems actually behave in production, including their failure modes.
  

• Candidates focused purely on traditional, deterministic software product work with no interest in model behavior, prompt design, or evaluation methodology.
  

• Candidates uncomfortable with ambiguity — AI product behavior is probabilistic, and specs need to account for that rather than assume fixed outcomes.