Pair: Open Tool for Coding with GPTs, Built by Coding with GPTs
Github Copilot is a great tool for leveraging GPTs while coding, but I find that it is too “open loop” for more complex tasks that require Q&A, feedback to guide it in a particular direction, iteration on code execution errors, etc. There is a large class of tasks that are better accomplished in an iterative, stateful chat-like interface. I have been experimenting with a local command line chat interface to GPT-4 and my mind was blown once again a few days ago when I copied documentation for a pretty involved API into the model context and managed to chat-guide GPT-4 to implement the API in…
In plain words
Pair is a command-line chat interface designed for iterative coding with GPT-4. Unlike GitHub Copilot's single-request approach, it enables back-and-forth conversation to tackle complex programming tasks, debug errors, and refine code through multiple rounds of feedback. The tool streamlines workflows by reducing manual copy-pasting between documentation, code, and AI responses, making it useful for developers who need guided iteration on involved implementations.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Github Copilot is a great tool for leveraging GPTs while coding, but I find that it is too “open loop” for more complex tasks that require Q&A, feedback to guide it in a particular direction, iteration on code execution errors, etc. There is a large class of tasks that are better accomplished in an iterative, stateful chat-like interface. I have been experimenting with a local command line chat interface to GPT-4 and my mind was blown once again a few days ago when I copied documentation for a pretty involved API into the model context and managed to chat-guide GPT-4 to implement the API in under 30 minutes, complete with a ridiculous amount of unit test coverage. This involved a lot of manual copy and pasting back and forth and other friction points that could be removed by a streamlined REPL interface optimized for code interactions. It occurred to me that it would be fun to build such a tool, and as the ultimate act of dogfooding, try to build it with GPT! So PAIR is the starting point here. You can see a recent commit message has a log of my interactions with the model that produced that commit. Next step is to add better mechanisms to manage the model input context (e.g. make it easy for the model to see the latest version of a source file when needed) followed by mechanisms for allowing the model to suggest changes via diffs that are quickly reviewed and accepted by the human in the loop before being applied to the file and tested. I would love to hear from others who have experimented with GPT pair programming in a chat-style interface and any feedback you might have on your experience with it.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, March 2023
the whole month →
- BI
I'm a big fan of the BBC podcast In Our Time -- and (like most people) I've been playing with the OpenAI APIs. In Our Time has almost 1,000 episodes on everything from Cleopatra to the evolution of teeth to plasma physics, all still available, so it's my starting point to learn about most topics. But it's not well organised. So here are the episodes sorted by library code. It's fun to explore. Web scraping is usually pretty tedious, but I found that I could send the minimised HTML to GPT-3 and get (almost) perfect JSON back: the prompt includes the Typescript definition. At the same time I…
AI · 2023 · genmon.github.io