Product Manager, Agents

About Luma
Luma's mission is to build unified general intelligence that can generate, understand, and operate in the physical world. We believe that multimodality is critical for intelligence. To go beyond language models and build more aware, capable and useful systems, the next step function change will come from vision. We are working on training and scaling up multimodal foundation models for systems that can see and understand, show and explain, and eventually interact with our world to effect change.

Where You Come In
Luma Agents is one of our three core product surfaces – alongside foundation models and Canvas – and the primary way enterprise customers will interact with our capabilities in production. Agents are how Luma's multimodal intelligence actually does work: executing multi-step creative workflows, operating within customer pipelines, making decisions that touch real brand assets, real campaigns, and real production deadlines. Building agents that enterprise customers trust to operate autonomously in high‑stakes creative and commercial environments is a fundamentally different product challenge than building a model or a canvas. It requires designing for safety, governance, and auditability as core product surfaces, not afterthoughts. It requires building evaluation frameworks that measure whether agents are actually doing what customers need, not just whether the underlying model is capable. It requires defining the primitives, protocols, and developer experience that let both our internal teams and external customers build agents that scale from experimentation to production. This is a 0-to-1 role. You will own the Luma Agents product end-to-end – the platform, the trust layer, the developer experience, and the agent capabilities that enterprise customers in marketing, advertising, and entertainment will deploy into their workflows. You'll work at the intersection of frontier research, product engineering, and real customer deployments to define what Luma Agents becomes.

We're looking for a principal-level product leader with deep experience building agentic AI products and the technical fluency to work directly with researchers and engineers on agent architecture, evaluation systems, and safety design.

What You’ll Do

Own the end‑to‑end product vision, strategy, and roadmap for Luma Agents – from agent infrastructure and orchestration through to the capabilities enterprise customers deploy in production

Define and build the trust and governance layer as a core product surface – audit trails, safety controls, permissioning, and the evaluation frameworks that give enterprise customers confidence to deploy agents autonomously

Design the agent primitives, SDKs, and developer experience that enable both internal teams and external customers to build, test, and deploy agents on top of Luma’s multimodal models

Work directly with research teams to translate frontier model capabilities – multimodal understanding, tool use, long‑horizon planning – into reliable, shippable agent behaviors

Build evaluation systems that measure agent performance in real‑world enterprise workflows, not just benchmark scores – defining what “good” looks like for agents operating in marketing, advertising, and entertainment pipelines

Drive cross‑functional execution across research, product engineering, forward‑deployed teams, and go‑to‑market within Luma’s pod‑based structure

Partner with the Enterprise PM and Forward Deployed PM to ensure the agent platform serves the needs of real customers and scales from bespoke deployments to broadly available products

Embed safety‑by‑design principles into all agent products – ensuring agents operate within defined boundaries, fail gracefully, and maintain the trust that enterprise customers require

Who You Are

Principal or Staff‑level product management experience (L6 seniority or above) with a track record of building agentic AI products – not just traditional SaaS or chatbot interfaces

Deep technical fluency in agent architectures, orchestration patterns, tool use, evaluation methodologies, and safety design – you are the PM that researchers and engineers want in the room

You’ve built products where trust, governance, and auditability were core to the product surface, not features bolted on later – you understand that in enterprise agent deployment, the safety layer is the product

You’ve shipped 0‑to‑1 products in high‑ambiguity environments and can define product direction when there is no existing playbook

You can design developer experiences, APIs, and platform primitives – you think about how agents get built and deployed, not just what they do

You operate with extremely high agency and a strong self‑starter mentality in fast‑moving, unstructured environments

You have a general manager mindset – you think about commercial viability and revenue outcomes alongside technical excellence

You can partner with and influence frontier research teams, translating emergent capabilities into product bets without waiting for capabilities to be fully mature

What Sets You Apart

Experience building agent platforms, agent infrastructure, or agent evaluation systems at a frontier AI lab or enterprise AI company

Familiarity with protocol and ecosystem design (MCP, agent SDKs, tool‑use frameworks) and the developer experience challenges specific to agentic products

Background in multimodal AI – you understand the unique opportunities and challenges of agents that can see, understand, and generate across modalities

Domain exposure in marketing, advertising, entertainment, or related verticals where agents operate in high‑stakes creative and commercial workflows

Experience at companies like Anthropic, OpenAI, Harvey, Sierra, Decagon, or Glean where agentic products are the core business

You’ve designed evaluation frameworks for agent systems in production – you know that benchmarks and real‑world performance are different problems

#J-18808-Ljbffr