Senior/Principal Product Manager

Say hello to opportunities. If you’re looking to be part of what’s next in communication, you’re in the right place. At RingCentral, we believe the best customer experiences happen when humans and AI work together. Our agentic voice AI portfolio—AIR, AVA, and ACE—brings together automation, assistance, and insights across the entire conversation lifecycle. The result? More seamless, intelligent experiences for businesses everywhere. With $2.5B+ in ARR and $250M invested in R&D annually, we’re building the future of AI-powered business communications. RingCentral is moving from AI that assists people to AI that completes work. Our AI-first customer engagement portfolio — AIR, AIR Pro, RingCX, and the AI agent capabilities we are shipping across voice and more than twenty digital channels — is being adopted at scale, with RingCX growing roughly 70% year over year and more than 1,700 business customers. We are turning individual AI features into a coherent agentic platform that enterprises can build on, govern, measure, and trust. That is what this role owns. You will define and ship the platform layer that lets an AI agent reason over a goal, call real business systems, respect enterprise guardrails, hand off to a human with full context, and be evaluated with the same rigor a contact center applies to its best people. You will do it at a company that already owns the voice channel, the enterprise relationships, and the distribution — which means your decisions reach production traffic quickly, and mistakes are expensive. We are looking for someone who has done this where it is hardest: enterprise conversational and agentic AI. If you have built agent platforms, agent builders, agent-assist products, or CX automation at a company competing in this category, you will recognize most of this description. What you will own Four surfaces, one product story. You will not own all four equally on day one — you and your leadership will sequence them — but you are accountable for how they fit together. 1. The agentic platform and agent builder Agent orchestration: goal decomposition, multi-step reasoning, and where the line sits between deterministic workflow and model-driven autonomy for a regulated enterprise buyer. Tool use and actions: how agents call CRM, ticketing, order management, and homegrown systems safely — schema definition, auth, retries, idempotency, timeouts, and failure semantics. Knowledge grounding and retrieval: connecting agents to knowledge bases, policy documents, and structured data, with attention to freshness, permission-aware retrieval, and source citation. Guardrails and control: topic boundaries, escalation triggers, human-in-the-loop checkpoints, approval gates for consequential actions, and PII handling. Agent lifecycle: authoring, versioning, simulation, staged rollout, rollback, and deprecation — with a no- code path for CX administrators and a pro-code path for developers on the same underlying model. Multi-agent patterns: supervisor and specialist agents, routing between agents, and shared context across a customer journey. 2. Developer platform, APIs, and ecosystem Public APIs, SDKs, webhooks, and event streams that make the agentic platform extensible by customers, systems integrators, and independent software vendors. Interoperability with the agent tooling standards our customers are standardizing on, including model- context and tool-interop protocols, and a clear position on where we adopt versus differentiate. Integration and connector strategy: what we build, what partners build, and what the marketplace surface looks like. Platform quality attributes treated as product requirements rather than afterthoughts: multi-tenancy, rate limiting, versioning and deprecation policy, sandbox environments, and identity and access management for both human users and non-human agent identities. Developer experience: documentation, quickstarts, reference implementations, and time-to-first-working-agent as a tracked metric. 3. Evaluation, observability, and model operations The evaluation system: offline test sets, conversation replay and simulation, model-as-judge scoring with human calibration, regression suites that gate release, and golden datasets built from real traffic. Production observability: transcript and trace inspection, tool-call success rates, hallucination and policy-violation detection, latency budgets, and drift monitoring. Model lifecycle: model and prompt selection, A/B and shadow testing, upgrade paths when a foundation model changes underneath us, and the cost-quality-latency tradeoff expressed as an explicit product decision. Unit economics: cost per conversation and cost per resolution, plus the levers — routing to smaller models, caching, context trimming — that move them without degrading outcomes. 4. Customer engagement outcomes The metrics the buyer actually cares about: containment and self-service resolution rate, first-contact resolution, average handle time, transfer rate, CSAT, and revenue outcomes for sales and retention use cases. Voice-native quality: latency and turn-taking, interruption and barge-in handling, ASR and TTS quality across accents, and graceful degradation on poor audio. Agent assist and the human-AI seam: real-time guidance, summarization, after-call work, and QA and coaching workflows built on conversation data. Multilingual and omnichannel parity, and the pricing and packaging implications of consumption-based AI agents. How you will work Write it down. You own strategy documents, PRDs, and decision records specific enough for engineering to build from and honest enough to survive a design review. Get in front of customers . You will run design partner programs, sit inside real deployments, listen to real call recordings, and bring back the difference between what customers say they want and what their operational data shows. Build with the bu