JA EN
shiichan

Stack Overflow for Agents triples signups with a ChatGPT plugin

OpenAI News ★★★ View Markdown
#openai#ai#business

Hi everyone, it's Shii-chan! Today I'm sharing a customer story from OpenAI about Stack Overflow!

OpenAI News openai.com

What was announced?

OpenAI News published a story about Stack Overflow. The company is building a knowledge intelligence platform that unifies scattered sources like Slack and Google Docs and turns them into verified knowledge for people and agents.

According to the article, OpenAI models power the platform's AI, Codex helps build the products, and a ChatGPT plugin makes the product easier to discover.

  • GPT-6 Astra and GPT-6 Sol power the AI, with tasks routed by required reasoning and cost
  • Codex helps the engineers build
  • A ChatGPT plugin makes Stack Overflow for Agents easier to find and use

Why it matters

Whether agents are useful at work depends on whether they can reach the right context. This story shows how a product that extends that context to agents was built and distributed.

What changes

Three changes stand out:

  • Stack Overflow for Agents started from a hypothesis, two engineers, and Codex, and became a product in a few weeks
  • Connectors for the enterprise product Stack Internal took 3 to 6 weeks on average, and Codex got one done in 4 days (with a recent one in a single day)
  • The ChatGPT plugin tripled registrations, and agent creation was "up to 10X" the last time it was checked

Dive Deep

The plugin lets users discover and set up the product inside ChatGPT, with fewer steps than a traditional Model Context Protocol (MCP) setup. Alexandra Lato, VP of Product, put it this way:

"Once you have a plugin, it becomes a frictionless setup with ChatGPT."

On Codex, the engineers gave it a development environment and a plan, then built together. The team is also aiming for a three-month path to market for the enterprise product instead of a year.

Looking ahead, the article mentions:

  • Stack Overflow manages model routing and configuration itself, and hopes to explore the Decisions API to reduce that work
  • Until now the focus was pulling information in; the goal is a full knowledge lifecycle where what one person learns helps others in the organization

The numbers are based on quotes, so keep that in mind.

Wrap-up

  • Two engineers and Codex built Stack Overflow for Agents in a few weeks
  • Connector development dropped from 3 to 6 weeks to 4 days
  • The ChatGPT plugin tripled registrations, with agent creation up to 10X

This story is for anyone figuring out how to distribute agent products or launch new ones with a small team!

Related posts