GenAI PM
tool2 mentions· Updated Nov 17, 2026

Teammate

A session-based tool used to build an open-world browser driving game and render a film. It is described as producing code-generated content without a game engine or external assets.

Key Highlights

  • Teammate was described as a session-based tool that can produce complex code-generated outputs beyond simple assistance.
  • Aravind Srinivas said Teammate was producing 11x more agent hours per week, up from 7x, with strong labor-value-to-token-cost economics.
  • A Teammate session on Fable 5.2 reportedly built a 9 MB open-world browser driving game with about 63,000 lines of JavaScript.
  • The showcased game was said to use a custom WebGL2 renderer with no game engine, no Three.js, and no external assets.
  • Teammate also rendered a 40-second film on a rented cloud GPU at 1080p60.

At a glance

Category
Code

Overview

Teammate is a session-based tool described in newsletter coverage as producing substantial autonomous work output and, in one notable example, building an open-world browser driving game and rendering a short film. The game was shared by Aravind Srinivas as having been created by a Teammate session on Fable 5.2, with code-generated content, a custom WebGL2 renderer, and no game engine, Three.js, or external assets.

For AI Product Managers, Teammate matters as an example of agentic tooling moving beyond text assistance into end-to-end production tasks. The mentions suggest two important dimensions: measurable labor leverage inside organizations, and the ability to generate complex software artifacts with tightly constrained technical choices. Together, those signals make Teammate relevant for PMs evaluating AI tools for engineering productivity, creative production, and agent-based workflows.

Key Developments

  • 2026-11-15: Aravind Srinivas said Teammate was producing 11x more agent hours per week, up from 7x, with the number of employees gaining 8+ or 40+ extra hours more than tripling in a week since September. He estimated $80M–$180M in annual labor value versus about $10M in tokens.
  • 2026-11-17: Aravind Srinivas shared an open-world browser driving game built by a Teammate session on Fable 5.2. The project featured a code-generated neon 1986 Miami, no game engine, no Three.js, no external assets, a custom WebGL2 renderer, roughly 63,000 lines of JavaScript, and a 9 MB footprint. Teammate also rendered a 40-second film on a rented cloud GPU at 1080p60.

Relevance to AI PMs

1. Benchmark agent productivity with business metrics. The reported jump from 7x to 11x agent hours per week, plus the labor-value-to-token-cost framing, gives PMs a concrete template for evaluating AI tooling in terms of throughput, cost efficiency, and internal ROI.

2. Assess end-to-end autonomy, not just copilot features. Teammate was described as completing a complex creative-technical project across coding and rendering, which is useful for PMs comparing tools that orchestrate multi-step execution versus tools that only assist with isolated tasks.

3. Use constrained builds to test model capability. The browser game example is especially relevant because it was reportedly built without a game engine, Three.js, or external assets. PMs can apply similar constraints in pilot projects to understand how well an agent performs under architectural, dependency, or asset limitations.

Related

  • aravind-srinivas: Source of both cited mentions, including productivity claims and the game/film showcase involving Teammate.
  • fable-52: The game project was described as built by a Teammate session on Fable 5.2, making it a key platform or model context for the showcased output.
  • threejs: Specifically mentioned as not being used in the browser game, highlighting the claim that Teammate produced the experience without relying on that common 3D web graphics library.

Frequently asked questions

What is Teammate?

A session-based tool used to build an open-world browser driving game and render a film. It is described as producing code-generated content without a game engine or external assets.

What should AI product managers know about Teammate?

Teammate was described as a session-based tool that can produce complex code-generated outputs beyond simple assistance. Aravind Srinivas said Teammate was producing 11x more agent hours per week, up from 7x, with strong labor-value-to-token-cost economics. A Teammate session on Fable 5.2 reportedly built a 9 MB open-world browser driving game with about 63,000 lines of JavaScript.

Newsletter Mentions (2)

2026-11-17
“#7 𝕏 Aravind Srinivas shared an open-world browser driving game built by a Teammate session on Fable 5.2, featuring a code-generated neon 1986 Miami with no game engine, Three.js, or external assets. The 9 MB game uses a custom WebGL2 renderer and about 63,000 lines of JavaScript; Teammate rendered a 40-second film on a rented cloud GPU at 1080p60.”

#7 𝕏 Aravind Srinivas shared an open-world browser driving game built by a Teammate session on Fable 5.2, featuring a code-generated neon 1986 Miami with no game engine, Three.js, or external assets. The 9 MB game uses a custom WebGL2 renderer and about 63,000 lines of JavaScript; Teammate rendered a 40-second film on a rented cloud GPU at 1080p60.

2026-11-15
“Aravind Srinivas shared that Teammate was producing 11x more agent hours per week, up more than 50% week over week from 7x, while the number of employees gaining 8+ or 40+ extra hours more than tripled in a week since September.”

#3 𝕏 Aravind Srinivas shared that Teammate was producing 11x more agent hours per week, up more than 50% week over week from 7x, while the number of employees gaining 8+ or 40+ extra hours more than tripled in a week since September. He estimated $80M–$180M a year in labor value against about $10M in tokens—a 9x–19x ratio depending on the job.

Stay updated on Teammate

Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.

Subscribe Free