AiGanakStop hitting your AI limits
Drop your Claude chat history. Get your AI Efficiency Score, see what is eating your usage, and copy the fixes that raise it.
Private by design. Your chats are never stored or shared, and no one at AiGanak sees them. Our privacy promise
Sample result from made-up chats. Drop your own export above to see yours.
–
Your fixes, biggest first
Next, from what we found
Apply a fix or two, then drop a new export in a week. The page remembers this score for you and shows your before and after.
No one at AiGanak sees your chats
Chat history can hold client names, contracts and personal details. The scan is built so that it stays yours.
Your file and your chats are not kept by AiGanak, on any server or in any database.
No one at AiGanak reads your chats, and they are not shared with anyone or used to train anything.
You never connect your Claude or ChatGPT account and never share a password or API key.
Only if you leave your email for the Fix Pack do we receive your score and counts. Never chat text, chat names or files.
Two minutes, no passwords
You download your own history from Claude and drop it here. We never ask you to sign in or share a key, and we never keep the file.
- In Claude, open Settings, then Privacy, then Export data.
- Wait for the email from Anthropic. It usually comes within minutes.
- Download the conversations file (a zip). If you get a manifest with several links, drop the manifest here and we show which ones to download. Links expire after 24 hours.
- Drop the zip on this page.
Coming next: ChatGPT (Settings, Data controls, Export) and Gemini (Google Takeout, Gemini Apps), and Claude Code logs with exact usage.
Measured from your own chats
Every point traces back to chats in your file, and only you see the result. The same rules apply to everyone, and any change to them is dated here.
- Chat length, 35 points. The longer a chat runs, the more each new message costs. Work done in very long chats costs points.
- Repeated requests, 20 points. The same request or one-word reply sent again and again, which could be one instruction or one list.
- Context hygiene, 10 points. Documents and tool output carried through a chat, compared with what was actually written.
- Model fit and plan fit need data a Claude export does not include, so they are marked as not measured and the score is scaled to 100 from the parts that are.
The API value prices your use at Claude Opus 5.5 list prices checked 9 Nov 2026. It is an estimate and can run low for heavy agent work.
AiGanak Scan questions
Can't find it? Press ⌘K and ask.
What does AiGanak Scan do?
It reads the chat history you export from Claude and shows where your usage goes: an AI Efficiency Score out of 100, how much further your plan could go, the API value of your use and the share spent re-reading long chats. Then it lists ready fixes, each with a prompt to copy and the steps, ranked by how much they raise your score.
Does AiGanak see or keep my chats?
No. The scan is private by design: your chats are never stored or shared, no one at AiGanak sees them, and they are not used to train anything. If you leave your email for the Fix Pack, we receive only your score and counts, never chat text, chat names or files.
Why do I keep hitting my AI usage limit?
Mostly because of long chats. Every reply re-reads the whole chat before it, so a message in a chat that has run for weeks can cost many times what it costs in a fresh one. Large documents and tool output kept in the chat, one-word replies in long chats, and using the top model for simple tasks add to it. The scan shows which of these applies to you and by how much.
How is the score worked out?
Three parts are measured from a Claude export: chat length (35 points), repeated requests (20) and documents and tool output carried in chats (10). The total is scaled to 100. Model fit and plan fit need data the Claude export does not have, so they are marked as not measured. Click the score to see each part and the chats behind it.
How accurate is the API value?
It is an estimate, priced at Claude Opus 5.5 list prices, of what the same use would cost through the API. For heavy agent work the real figure can be higher. Read it as a measure of how hard your use pushes your plan, not what you paid.
How do I export my Claude history?
In Claude, open Settings, then Privacy, then Export data. You get an email with download links. Download the conversations file (a zip) and drop it on this page. If you receive a manifest file with several links, drop that and the page tells you which ones to download.