Director of Product Management, CS/AI

About the role

We’re hiring a Director of Product for AI & Agentic Customer Support to own the strategy and outcomes for how Flex resolves customer issues - automatically, intelligently, and at scale. Customer support at Flex is a multi-channel operation spanning AI voice, AI chat, live chat, email, and live human voice, serving bill payers managing payments, billing changes, onboarding and identity, account issues, and more. With hundreds of human agents and a rapidly growing bill payment base, the mission is to resolve more, faster, and with less friction.

This role owns the full agentic stack: the in-app customer experience that routes bill payers to the right channel and drives toward self-resolution; the LLM-native orchestration layer that powers automated resolution; the ML-driven intelligence for routing, personalization, and issue prediction; and the capability primitives that make human agents dramatically more effective. You will set the strategy, define the architecture, and be accountable for the outcomes — automation rates, customer resolution rates, first contact resolution, and average handle time.

This is a 0-to-1 role. The agentic architecture is being defined now, and you will define it with us. This role is ideal for someone who has built agentic or AI-driven systems in data-rich environments, knows how to write PRDs/specs that ML engineers (and their agents) trust, and has the judgment to choose LLM-native orchestration when it’s warranted and deterministic automation when it’s not.

What you’ll do

• Own the strategy and outcomes for all things agentic. Set the multi-quarter roadmap for automated resolution, intelligent routing, and agent augmentation. Define and own the key metrics — customer resolution rate, automation rate, first contact resolution, and average handle time reduction for human agents — and drive the product decisions that move them.
• Design the in-app customer support experience. Own how bill payers discover, access, and navigate support inside the Flex app. This includes intelligent channel routing — directing customers to the right channel (AI chat, voice, human escalation) based on issue type, history, and predicted resolution likelihood — as well as the in-app AI-driven interaction experience itself. The goal is maximum self-resolution before a human agent is ever needed.
• Build the agentic resolution platform. Design the LLM-native orchestration layer for fully automated CS resolution — tool-calling agents, multi-step workflows, context-aware decision trees, and graceful fallback routing to human queues. Define what actions agents can take (payment retries, balance corrections, account updates), what guardrails govern them, and how confidence thresholds drive escalation.
• Own the customer intelligence layer. Write PRDs for the ML-driven capabilities that power personalization, intent classification, dynamic routing, and proactive issue detection. Define the customer context model — what signals matter, how they are assembled, and how they are served in real time to both automated and human-facing systems.
• Drive the interaction data model. Work with the data platform team to establish the interaction ledger, event schema, and feedback loops that make sustained intelligence improvements possible at scale.
• Define the vendor integration product layer. Own requirements for how Flex’s AI voice and chat vendors integrate into the orchestration architecture — API contracts, context payloads, handoff protocols, and performance standards. You are not selecting vendors; you are ensuring the vendor layer is a composable component, not a constraint on what the agentic system can do.
• Enable AI assist for human agents. Partner with the CS product PM who owns the agent-facing UX to define the underlying capability primitives: suggested responses, real-time context surfacing, post-contact summarization, and escalation briefing. You own the intelligence layer; they own how agents experience it.
• Set the architecture direction. Author the decision records and strategy memos that inform multi-quarter investment across engineering, ML, and data platforms. Represent the AI & Agentic roadmap to Platform and CS leadership.

Key qualifications

• 6+ years of product management experience, with at least 2 years working directly on AI, ML, or agentic systems in a production environment.
• Demonstrated 0-to-1 track record: you have taken an AI-powered capability from concept through launch in an ambiguous, resource-constrained environment.
• Experience owning end-to-end customer-facing AI experiences, including how users are routed, how AI-driven interactions are designed, and how resolution quality is measured and improved.
• Ability to write PRDs for ML systems - feature engineering specs, model evaluation criteria, training data requirements, latency and accuracy tradeoffs. You do not need to write the code; you need to write documents that ML engineers trust.
• Hands-on experience with LLM-native orchestration: tool-calling, multi-step agents, prompt engineering in production contexts, context and memory management.
• Strong judgment on when to use LLM reasoning vs. deterministic rules - and the intellectual honesty to apply that judgment even when LLM is the more compelling answer.
• Comfort owning outcomes, not just output. You track the metrics that matter - resolution rates, automation rates, handle time - and treat them as your own, not your engineering team’s.
• Excellent written communication. At Flex, the PRD is the primary artifact of product leadership. Yours should reduce ambiguity, not create it.
• Bonus: Experience in fintech, payments, or any domain where automated actions carry financial or compliance risk. Prior exposure to contact center platforms, AI voice or chat vendors, or CRM integration patterns.