Agent in the empty chair: Amazon vets launch Primitive Labs, using AI to model customer behavior
Rohit Talluri learned a tradition at Amazon: always keep an empty chair in the room to represent the customer – a reminder of the people who will use whatever is built in the end.
Now, with AI coding tools making software faster than ever, Talluri and his co-founders, Amazon veterans Jean Farmer and Gabriel Fong, realize that the customer can easily be forgotten in the process. So they are creating a seat at the table for AI agents.
Lowo umbono ngemuva kwama-Primitive Labs. The startup builds what it calls behavioral intelligence: systems that look, think and act as customers do across software platforms and devices, helping product teams learn how people will react to a new feature, marketing or marketing decision before it’s shipped.
Traditional user research and focus groups can take weeks or months, so teams under pressure to ship quickly are tempted to skip them. Primitive Labs conducts such research with agents that mimic human behavior, aiming to make it a common step in software development.
“It puts people back at the center of the world created by AI,” Taluri said. “That’s the goal here.”
The goal, according to the startup’s presentation, is “to make human behavior the first priority of software development.” Lokho ugqozi lwegama le-Primitive Labs. The idea is to build products that people will understand, trust and continue to use – not the average user, but specific types of users in specific situations.
The founding team: Talluri, CEO of Primitive Labs, is joined by co-founders Farmer, CTO; and Fong, COO.
Fong and Taluri have worked together since 2020. At AWS in Seattle, Fong held product marketing and business account roles, then led sales and marketing at cloud consultancy DoiT International.
At Primitive Labs, his role is broader than sales and marketing, encompassing product direction, customer development and operations. Talluri describes him as very professional and involved in the core product work of the company.
Farmer and Taluri worked together at AWS on large-scale machine learning infrastructure, including the SageMaker HyperPod training service, before both moved to Amazon’s AGI organization.
Farmer worked on the Amazon Nova model’s ability to use software tools – designing how models call tools and take actions, and build systems to test and measure how well the resulting agents are performing. That work included benchmarks for the Model Context Protocol (MCP), an emerging standard for connecting AI models to external tools and data.
Talluri joined the AGI Autonomy Lab, a team Amazon assembled around talent hired from Adept, a San Francisco startup building AI agents that use the software themselves.
Amazon brought in Adept CEO David Luan, a former OpenAI executive, and other co-founders in 2024, and licensed the startup’s technology, putting Luan in charge of the lab. Talluri worked there on computing agents and helped launch the Nova Act, Amazon’s model for computing.
Taluri said he initially came close to leaving Amazon in 2025 to start a company, before its leaders directed him to Autonomy Lab to work under Luan (who has since left Amazon).
Sponsorship: Primitive Labs raised a pre-seed round, led by a16z Speedrun and joined by several small, innovative funds and a group of angel investors. Inkampani ayilivezi inani loxhaso.
Its launch lists backers including Olive Tree Capital, Cloverfield Fund and Unexpected Funding (from former TechCrunch editor Josh Constine), as well as angels such as Luan, Harsh Patel and Artur Kiulian, and others with backgrounds in OpenAI, Amazon, Google DeepMind, Databricks, Nvidia and Meta.
Primitive Labs will join the a16z Speedrun batch starting this month, and expects to raise the next round towards the end of the program, in September or October.
Headquarters: The company is based in San Francisco, where it works part-time at a16z’s Speedrun location, with plans to get its own office after making its first hire.
Taluri, a University of Washington graduate who read GeekWire as an intern and dreamed of starting his own startup, said the choice came down to the concentration of talent in San Francisco and the pace of AI research there, as well as the existing Speedrun program.
Primitive Labs posted its first job listing last week – for creative engineers, researchers and interns, in San Francisco or New York.
Product status: The company raises upfront capital and works with a small group of early customers who test its product and help refine it, including private previews with what Talluri described as Fortune 500 and Fortune 50 consumer technology and e-commerce brands.
The company plans to launch its products generally later this year.
How does this work: Agents work across devices including computers and phones, currently focusing on digital products and customer journeys. The company says it has also tested using them to measure reactions to physical products, such as product and packaging.
The basic research draws on integrated cognitive science, continuous learning and custom memory systems that modulate how people store information — work Talluri said the company plans to publish again in an open-source section in the coming months.
While other startups are working on agent-based simulation and automated testing of user interfaces, what sets Primitive Labs apart, says Taluri, is the focus on human interaction. That means build agents faithfully represent the users of a particular product, making that a standard layer of how software is built. He defined the important measure as moral fidelity, or how closely the agent’s choices track the person’s decisions.
Asked if the startup would keep an empty seat when it found an office, per Amazon’s tradition, Taluri didn’t hesitate. “100%,” he said. And yes, he said, they would be seeing an agent sitting there.
