01
Your Python app
Install the lightweight SDK and call a familiar client from any script, notebook, or service.
pip install pairitA local bridge for curious builders
Put your Python
in the loop.
PairIt lets your scripts talk to AI through the browser you already use. Bring your own ChatGPT, Claude, or Gemini session, or run a small model right on your machine.
No API keys for browser sessionsLocal models stay local
from pairit import Client
client = Client("chatgpt")
response = client.chat(
"Explain this to me"
)
> connected · response ready
⌘ Python
◌ Browser session
✦ 100% local
+
×
One APIfor browser AI and local models
Zero telemetryfor stateless inference
WebGPU readywith WASM fallback
Python firstbuilt for real workflows
The simple version
Three pieces.
One smooth handoff.
PairIt keeps the complicated part between your script and the browser, so you can focus on what you want to build.
02
Local bridge
The SDK and extension meet over a local WebSocket on 127.0.0.1:8765.
Connected locally
03
Your AI surface
Route to an authenticated browser tab or a model executing directly inside Chrome.
ChatVisionEmbed
Choose your path
Same client.
Different kind of private.
Start with the workflow that matches the job. You can use both in the same project.
STATEFUL↗
Borrow the browser
Use an AI tab you are already signed into. Your code sends prompts through the open session, with no provider API key to manage.
ChatGPTClaudeGemini
Best for rich, current browser sessions
STATELESS↗
Keep the model here
Download a supported model once, then run chat, vision, or embeddings locally in Chrome using WebGPU or WASM.
Best for offline and sensitive data
Your first run
From zero to
first response.
Get the extension, open a session, and let your Python code do the talking.
Read the full docs01
Install the bridge
Install the Python package in the environment where your code runs.
pip install pairit02
Connect a surface
Install PairIt, pin it, then either open ChatGPT, Claude, or Gemini, or download a model in the popup.
03
Make the call
Point Client at the model or provider you want. The API stays pleasantly small.
hello_pairit.py
from pairit import Client client = Client("chatgpt") response = client.chat("What should I build today?") print(response.content)
⌂
Built with boundaries
Your data has a clear address.
Browser sessions stay in your browser. Stateless prompts, images, and embeddings run inside Chrome and do not leave your machine. Downloads are verified before they are marked ready.
01 Local bridge02 No credential export03 SHA-256 checks
Ready when you are