Local-first · zero uploads · works offline
Every conversation, inspected in your browser.
Drag in a ChatGPT, Claude, Gemini or DeepSeek export — conversations.json, Gemini_History.json, .jsonl or the whole ZIP — and unlock secret scanning, custom instructions, full-text search, cost estimates and activity heatmaps. Parsed with JSZip 100% in your RAM.
Drop your export here
Drag & drop an AI chat export — ZIP, conversations.json, Gemini_History.json, chat_history.json or .jsonl — or click to browse. You can also paste files with Ctrl+V.
Parsed in your browser's RAM — zero bytes leave this device
JSZip decompression runs fully in memory.
ChatGPT, Claude, Gemini, DeepSeek, Llama, Qwen — and any role-based JSON/JSONL chat export.
No account, no upload, no telemetry — ever.
How it works
Three steps to total visibility.
From an unreadable JSON dump to a fully searchable, exportable archive — without ever sending a byte to a server.
Request your export
ChatGPT, Claude and Gemini all let you download your history from account settings. DeepSeek, Llama and Qwen exports work too — any role-based JSON or JSONL.
Drop it here
JSZip decompresses the archive entirely in your browser memory. Nothing is uploaded — the file never leaves your device.
Analyze & export
Scan for secrets, isolate custom instructions, search every thread and export to Markdown, clean text or JSON.
Why client-side parsing is the only private option
Your prompts are your secrets. Keep them that way.
Cloud tools that "sync" your history route your prompt logs through their servers — which means your confidential code, business strategy and personal data pass through infrastructure you don't control. AI Data Inspector is the opposite architecture:
-
No network calls, ever
The analyzer never connects to any AI provider or to us. Pull the network cable and it keeps working.
-
Decompression in RAM
JSZip expands archives in your browser heap; parsing and search run on a background thread on your own CPU.
-
Nothing is stored
No accounts, no telemetry, no tracking by us. Close the tab and every trace is gone — only your explicit consent choice and theme preference remain.
-
Consent-first ads
Marketing pages may show ads; they only set cookies if you opt in, and they never touch your conversation data.
const promise = parseInBrowser(zip);
// file.files → JSZip.loadAsync → RAM heap
await promise;
// nothing was POSTed · nothing was stored
fetch('/api/upload'); // ⇦ never happensFAQ
Questions, answered.
The short version: private, offline, free, and it works with almost any AI export.
Is my chat data uploaded to a server?
No. Every archive is decompressed and parsed entirely in your browser RAM with JSZip. Nothing is sent to any server — pull the network cable and it keeps working. Close the tab and every trace is gone.
Which AI platforms are supported?
ChatGPT (conversations.json), Claude, Gemini (Gemini_History.json), DeepSeek, Llama, Qwen — plus any role-based JSON or JSONL export, for example from Grok, Mistral or a custom pipeline.
How do I export my ChatGPT or Claude history?
ChatGPT: Settings → Data Controls → Export Data, then grab the ZIP from your email. Claude: Settings → Account → Export Data. Gemini: Data and privacy → Download your data. Then just drop the file into the analyzer.
What can I do with my export here?
Full-text search across every thread, a PII & secret scanner, custom-instruction isolation, cost estimates, activity heatmaps, and exports to Markdown, clean text or JSON.
Can it find exposed secrets in my history?
Yes. The scanner flags API keys, passwords, tokens, emails and credit-card-like patterns across all conversations, so you can redact or delete them before sharing or backing up.
Is AI Data Inspector free?
Yes — free forever, no account, no sign-up. There is an optional one-time tip that removes ads and unlocks the Pro Supporter badge, but the core analyzer never costs anything.
Does it work offline?
Yes. The whole analyzer runs locally in your browser; after the page loads it works fully offline with no network calls.