
I use AI daily and kept hitting the free tier limit way too fast. Got annoyed enough to actually build a fix.
I made Lakon - a browser extension that adds a small button right inside Claude, ChatGPT, and Gemini. Click it before sending, and it compresses your prompt down to only what the model actually needs.
No tab switching. No copy‑pasting. Just click and send.
Why it works: LLMs don’t read prompts like humans. They pay most attention to the first and last few words. Everything in the middle (“I was wondering if you could help me…” / “thanks so much”) is basically invisible to the model but still counts against your quota.
Example:
Raw Prompt: 77 tokens
Compressed Prompt: 17 tokens
→ Same answer back.
The extension is free, installs in ~2 minutes directly from the site, no account needed. There’s also a web version if you want to try it without installing.
Demo + install: Lakon Web