Skip to main content
Memory is a short list of notes and files that Vowen indexes once and then gives to the AI as background context when you use Command Mode. Use it for facts you would otherwise repeat, such as your address, a credential ID or your current projects. The Free plan holds 3 Memory items. Pro removes the limit.

Where to find it

Click Language models in the sidebar, then the Memory tab. The page has two tabs: Configuration and Memory. Language models › Memory tab with a note and two files marked Indexed The banner at the top reads Store essential files for smarter, context-aware responses. with the hint Hold to ask questions and your memory files will be used for context., where the hotkey is your Command Mode shortcut.
Memory needs an AI provider. Without one, the tab shows a lock card with a Connect an AI provider button that opens the Configuration tab. Indexing is a model call, so nothing can be indexed until a provider is connected.

Where Memory is used

To change how your dictation is cleaned up, use Custom Instructions or a Tone instead.

How to add a note

1

Click Add Note

The dialog has two fields, Name and Content.
2

Fill it in and save

Click Save Note. Indexing starts on its own.
3

Wait for Indexed

The row’s status changes to Indexed when the item is ready.
Good candidates for notes:
  • Reference numbers, such as a passport, tax ID or frequent flyer number
  • Certifications, with credential IDs and dates
  • Your name, role, company, mailing address and signature block
  • Current projects, team members and tech stack

How to add a file

Drag a file onto the banner, click the banner to browse, or click Upload File. You add one file at a time. Supported types: .txt, .md, .pdf, .docx, .json, .csv, .xml, .html, .htm.
  • Text formats (.txt, .md, .json, .csv, .xml, .html, .htm) are read directly.
  • Documents (.pdf, .docx) are sent to your AI provider’s document API. Only Google Gemini and Anthropic can read them, and Anthropic reads PDF but not DOCX. With any other provider the item ends in Error.
Vowen copies the file into its own storage when you add it. Moving or deleting the original does not affect the Memory item, and editing the original does nothing until you re-index.

How indexing works

Vowen sends each item to your AI provider once and asks it to extract the lasting facts: personal and professional details, dates, numbers, preferences and requirements. That extract is what Command Mode receives, not the original text. Very long content is shortened before indexing. Only Indexed items reach the AI. Each row has Index this item (or Re-index this item once it has succeeded) and Delete this item. Re-index an item after fixing whatever made it fail. There is no edit option. To change an item, delete it and add it again.

Free plan limit

The Free plan holds 3 items. Notes and files count toward the same total. The counter next to Upload File shows how many you have. If you had more than three items on Pro, the extra items stay visible with a Pro tag and are no longer used as context. Click the tag to see upgrade options. Free plans also include one custom utility built from a template, so Memory can reach a utility on the Free plan too.

Memory stays on this device

Memory is not included in Cloud Sync or in Import & Export Data. On a new computer, add your items again.

Example

  • Memory note AWS certification: “AWS Certified Solutions Architect Associate, credential ID ABC-12345-XYZ, earned 2024-06-12, expires 2027-06-12.”
  • Command: “Update my LinkedIn About section with my latest AWS certification.”
  • Result: the credential ID and dates come from Memory.

Tips

  • Keep it short. Every indexed item is sent with every Command Mode request, so a large Memory adds latency and cost to requests that don’t need it.
  • Prefer notes to long documents. A five-line note indexes more precisely than a long PDF that gets shortened.
  • Use clear names. The name is sent to the model with the extract.
  • Don’t store passwords, API keys or card numbers. Indexing sends the content to your AI provider, and documents are uploaded in full.