Your personal and business data hub

Your information is scattered. Give it a home.

Email accounts, documents, notes, calendars, and past AI conversations can all hold part of the answer. Bring the sources you choose into your Talavine data hub so you spend less time gathering and explaining them again.

Start with the sources that matter to you.

Email and messages

Connect supported accounts and channels. Keep the conversations that explain your relationships and work close to the questions you ask.

Notes and preferences

Write down what matters. Review your About Me note and preferences yourself. Secret notes stay out of AI retrieval.

Documents and calendar

Bring in supporting files and connected calendar information. Find the history behind a deadline, meeting, or decision.

Past AI conversations and exports

Import a ChatGPT data export, or readable text from other sources through the general archive importer. Supported formats vary; importing a transcript does not copy a service's private internal memory.

Bring your ChatGPT conversations, too.

  1. Get your exportIn an eligible ChatGPT account, open Settings → Data controls → Export data. Download the ZIP when it is ready. See OpenAI's export instructions and availability.
  2. Choose the source in TalavineOpen Imported Sources, add a ChatGPT data export source, and select the ZIP or conversations.json. Conversations become searchable Notes with speaker labels and dates. Supported attachments are preserved locally.
  3. Let Talavine make it searchableThe background indexer uses your configured AI to identify source-linked mentions of people, projects, companies, and topics. Local, CLI, and cloud routes can process small sections; progress resumes after restart. The Notes and graph stay local. Review progress and open the original Notes from Imported Sources.
  4. Refresh when you have more historyImport a newer export manually, or enable Automatic file refresh for one local export file. Replace that file with each new export; Talavine checks it while running and skips identical imports. It doesn't sign into ChatGPT or request exports for you.

Search can become useful before graph processing finishes. Processing needs a configured, eligible AI model; it waits if one is not ready. CLI and cloud providers may receive excerpts during background indexing as well as when answering questions. Imported statements remain source material, including AI replies and hypothetical questions; they don't automatically become confirmed facts about you.

Your data stays yours. Your AI can change.

Your Talavine profile and communication memory live on your machine. Cloud models are optional. When enabled, relevant request context is sent to the selected provider under the configured routing and privacy policy.

Choose a supported model for your conversation without moving your profile. Different models may give different answers; your stored sources remain available. You can inspect imported Notes, delete them, and export selected knowledge-map material as Markdown. Export a profile to move it into another Talavine installation; that backup is not a universal import format for other AI apps.

Imports stay on the current device by default. Sharing imported text with paired desktops is a separate choice; preserved archive attachments remain on the importing device. Removing a source registration keeps its Notes, so delete unwanted source material in Notes as well.

Explore knowledge and source-backed answers →

Agents can help keep information coming in.

Use a collector from the Agent Library for supported signed-in AI accounts, or give a research helper a specific job with the tools you connect. Review saved information and its source. Account access, collection coverage, and schedules depend on the configured agent.

How information collectors work →