Founding Product Manager — Agentic Application

Seed stage
· Remote (Hong Kong / Shenzhen / Guangzhou preferred)
· Reports directly to the founder 1. About Gro AI Founded by the former CDO of Huawei Cloud Asia Pacific, Gro AI is building next-generation AI Sales Infrastructure. Mission: Build the best way to sell with AI. Our customers: leading tech startups and listed companies globally Website: https://thegro.ai/ 2. Why this role matters This is Gro AI's first core product role. Gro is an autonomous agent platform: self-driving sales, governed by you. Agents produce most of our specs, builds and tests: fast, but prone to drift from what users need. As a founding team member reporting to the founder, you are the human judgment in that loop: you define what good looks like, design the UI and UX, run QA and take each release from spec to customer. You need to show: you have shipped an AI agent or LLM product to real users and owned its quality after launch, including a QA process you designed yourself. 3. What you'll own
• PRD: own Gro's PRD (agent specs, autonomy levels L1–L5, glossary, skills registry) as the single source of truth.
• Bridge CEO and dev team: turn the CEO's intent into testable requirements for human and agent developers, and ensure what ships matches it.
• QA and acceptance standards: define the quality bar and own QA from test plan through UAT to sign-off. Block releases that miss it.
• Evals: build eval sets and keep the AI judge calibrated to human judgment.
• Defect management: triage failures, route them to owners, verify fixes and turn recurring issues into platform rules.
• UI and UX design: own user journeys, wireframes, prototypes and developer-ready UI specs. Make agent status, approvals and error recovery clear.
• Discovery and usability: test with pilot customers, decide with the CEO what to build next, and verify the shipped UI matches the design. 4. What we're looking for Agentic product skills
• Evals: you write test cases and graders for non-deterministic outputs (pass/fail, code assertions, LLM-as-judge) and read traces.
• Human-in-the-loop design: what needs approval, what runs alone, how agents escalate.
• Technical fluency: prompting, context, tool use and MCP, retrieval, model choice, and cost and latency tradeoffs, deep enough to argue with engineers; plus SQL.
• Unit economics: you estimate cost per outcome and spot money-losing features. UI and UX design skills
• Hands-on product design: you have designed and shipped web app interfaces, from user flows to high-fidelity UI, and can explain your decisions.
• Tools and handoff: Figma or equivalent for prototypes and developer-ready specs, including components, states and responsive behaviour.
• Agentic UX: you make clear what the agent is doing, what needs approval and how to recover.
• Visual craft: clear hierarchy, layout, typography and accessible patterns. QA and delivery skills
• QA

experience:
you have owned test plans, test cases, regression suites, UAT and release sign-off.
• Defect lifecycle: bug triage, severity rules, root-cause analysis and fix verification.
• Standards from judgment: you turn tacit judgment into rubrics and examples that engineers and AI judges apply consistently.
• End-to-end delivery: releases to production with release notes, known limits and rollback plans.
• Product taste: you judge agent output as a user would and explain why it is good or bad. How you work
• AI-native: you use Claude, Cursor or similar agents daily and review their output critically.
• High agency: you turn vague problems into shippable plans and set scope, priority and cadence.
• Say no with evidence: you align the team on one definition of done and hold the line when a release isn't ready.
• Clear writing and customer empathy: you write precise documents and work directly with early customers on pilots.
•

Experience:
4+ years in product management or equivalent product and QA work; fluent English. 5. Nice-to-haves
• Sales tech experience, e.g. Clay, ***, Apollo, Outreach or Gong.
• Test automation or eval tools (Playwright, Cypress, LangSmith, Langfuse, Ragas).
• LinkedIn and email deliverability, platform limits and compliance.
• Outcome
- or usage-based pricing; multi-agent products.
• Reading code or prototyping yourself.
• Mandarin Chinese. 6. Why join us
• Work directly with the founder: Leo served hundreds of enterprise customers as CDO of Huawei Cloud Asia Pacific.
• Learn from a multi-agent systems expert: co-founder and Chief Science Officer Dr. Owen Zhao was an Agentic AI researcher at Huawei's Noah's Ark Lab and founding AI engineer at Final Round AI, specializing in multi-agent systems.
• Build with an agent swarm: direct and judge the agents in our autonomous dev factory that spec, build and test.
• Real-world impact: what you ship reaches global customers fast, and you set how Gro defines and ships quality.
• Competitive rewards: competitive salary and early equity.