דרושים » מוצר » AI Product Manager

We are looking for a Senior Product Designer who designs products and builds the tools that change how design gets done.

This role has two halves. You will own the design of a core area of the platform, end to end, at the same craft bar as everyone else on the team. Alongside that, you will build the agents, skills and internal tools that reshape how our whole design team works., and prove them on your own product area first.

We think these belong together. Someone rebuilding the workflow should be living inside it every day, so what they build survives contact with real product work. If you have been quietly automating your own design process and wishing it were your actual job, this is that job.

We already have an AI agent in the product. It works, it is early, and it is nowhere near what it should be. How far it goes, and what it grows into, is genuinely open. The designer who understands agents best will have a real voice in answering that.

This is an individual contributor role. You will mentor, and you will change how the team works, but you will not manage people.

Key Responsibilities:

Own a core product area: Take full design ownership of a significant surface, from discovery through shipped experience and beyond launch, working directly with product and engineering.
Build the agentic design toolchain: Create and maintain the skills, agents and commands the design team uses day to day, covering the path from ticket to reviewable prototype.
Prototype in working code: Move from problem statement to a clickable, testable artifact quickly by orchestrating agents to do the building while you review, steer and cut.
Orchestrate agents through a product cycle: Decide what gets delegated, what stays human, and where the human gate sits, then write it down so others can repeat it.
Make new ways of working stick: Run the rituals that turn tools into habits. Prototyping sprints, office hours, pairing, internal demos, and the documentation behind them. Retire what does not earn its place.
Keep us honest about quality: Define how we know an agent output is good, build lightweight checks for what you ship, and tell us plainly when the agentic path is worse than the manual one.