Project T900 Autor: Project T900

Give AI the right context, not all your context.

When you work with ChatGPT, Claude or Gemini on real material (notes, briefs, documents, past decisions), pasting everything in buries what matters. Project T900 prepares the context for you with the T900 Engine, on your device, and shows you every choice before anything reaches the AI.

WHAT IT DOES
- Works in chats, like your AI tool: each chat keeps its own memory, and a follow-up leaves out what the chat already sent, so you send only what is new.
- Keeps what is relevant to your task and leaves out repetition and noise, with the reason for each passage it keeps.
- Protects the details that carry weight: dates, figures, quotes, prohibitions and constraints.
- Flags statements that contradict each other, such as an old launch date next to the new one, and lets you decide which to keep.
- Remembers what each chat should keep in mind (a dietary need, a decision, an approved claim) and suggests it, most relevant first. Export a chat's memory as a PDF, made on your device.
- Starts with what your other AI assistant already knows about you: copy a prompt into it, paste its answer back, and keep the facts you want. They go into the pack on their own when they apply, where you can see them and take them out before you press Send.
- Reads PDF, Word and Markdown files, and printed English text in images and scanned PDF pages, on your device.
- On ChatGPT, Claude or Gemini, prepares the prompt you are writing, in the chat you pick, and puts the context above it as soon as it is ready. Take it out again while the popup is open. You always press Send yourself.
- Starts simple. Turn on Nerd mode for the advanced options: projects and workspaces, optimization modes, word budgets, relevance scores and token estimates.

PRIVATE BY DESIGN
- No account, no servers, no analytics, no tracking.
- Project T900 makes no network requests of its own, and its pages are locked down so the browser would refuse any.
- Your library stays in your browser.
- It looks at a page only when you open Project T900 on it, and reads only your prompt, or text you selected, when you click.
- Nothing is ever sent for you, and it never presses Send. You agree once how copying and inserting work (or choose a tick on every pack), and a pack holding a secret you added back asks first.

HONEST ABOUT LIMITS
The T900 Engine is not an AI model: it selects and orders your own text, and never writes or rewrites any of it. Project T900 does not guarantee better answers, prevent hallucinations, or promise token, speed or cost savings, and its token counts are local estimates. What it does is make the choice of context visible and reviewable, so you decide what the AI sees.