How it works

Everything Prophetica Memory does today, in plain terms.

What it is

One memory that every AI you use can read from and write to.

Tell one assistant something once and the others know it. Each assistant saves short notes, called memories, as you work, and any of them can search those notes later. Prophetica Memory also maps how your memories connect.

What goes in

It fills as you work.

Anything you ask an AI to remember. Plans, from the first idea to the booking. People, dates and what they told you. Research, notes and feedback. Decisions made in past coding sessions with Claude Code and Codex, read in the background. Email, chat and documents from the tools you connect, each linked to its source.

Works with

The assistants you already use.

Claude on web, desktop and mobile. Claude Code. ChatGPT. The Gemini web app, and the Gemini mobile app once you have connected on the web. Codex. Any app that can add an MCP server. ChatGPT needs a paid plan that supports custom MCP servers. Google limits custom apps in Gemini to people aged 18 or over in the US with a personal Google account.

Connects to

Connect a tool once.

Gmail, Slack, Google Drive and Notion connect today. New items become memories that link back to where they came from. More are in development, including Outlook, Obsidian, GitHub and Linear.

Setting up

A few minutes. Nothing to install.

Add your Prophetica address as a connector or MCP server in your assistant. Sign in on the page that opens. There is no API key. Start a new chat, and your assistant saves what matters and searches your memory when you ask. You can also paste one setup sentence into your assistant and let it walk you through.

Remember

Say remember.

Say remember in any assistant and it is saved for all of them. Each memory records which assistant it came from.

The map

A map of everything you know.

Each memory is read for the people, projects, tools and ideas in it. Names already known join their place in the map, and new ones are added. Select anything to see what is known about it and what it connects to.

Pages

A page for every person, project and tool.

Every line on a page cites its source. When something changes, the page is updated and points to the new source, so your assistants always read the latest version.

History

Every change is versioned.

You can see which assistant made each change, and roll back to any earlier version.

Sharing

Share a page like a document.

Add someone by email as an editor and their AI can add to the page. Turn on a link and anyone with it can view. When a friend shares a space with you, your AI can search it alongside your own memory, for one question or for every search. Nothing is visible to anyone unless you share it.

Efficiency

Fewer tokens than files.

Assistants search memory and usually find the answer in the first search, where reading files means opening them one after another. In our own tests, assistants used up to 60% fewer tokens.

Privacy

Private unless you share.

Only the apps you approve and Prophetica’s own processing can open your memories, and every access is checked and logged. AI models receive only what the task needs. Your memories are never sold and never used to train AI models. Everything is encrypted in transit and at rest. Disconnect an app in account settings and it stops working within seconds. Export everything, delete one memory, or delete your whole account at any time.

Nothing to build

No hardware. No side project.

Building this yourself means a server or GPU, a database, an embedding model, scripts to wire them together, and the upkeep that follows. With Prophetica Memory you sign in and connect your AI.

Beyond the second brain

No filing. No template.

The second brain method asks you to take notes and organize them by hand. Here your assistants save notes as you work, the structure builds itself, and any of them can search it.