AI processing and your content
Several Multilo features chat, agents, and inline suggestions work by sending the content you choose to act on (your prompt, selected text, document context, and any images you attach) to an AI model provider that generates a response. This only happens when you trigger the feature.
Content is sent over an encrypted connection to the model provider for that request (currently Anthropic, OpenAI, and/or Google, depending on the model you pick).
We send the minimum context the feature needs. We do not send your file names, project paths, or IP address to the model provider.
We instruct providers not to retain content for training where that control is available to us (for example, requesting no-storage handling on supported endpoints). Providers may retain limited data briefly for abuse monitoring under their own terms.
AI output can be wrong or incomplete. You are responsible for reviewing AI-assisted work before you rely on, submit, or publish it.
Sending content to a model provider is required for these features to function. If you do not want your content processed by an AI provider, do not use the AI features; the rest of the app still works.
The specific providers we use can change as we add or remove models. The active list of model providers is reflected in the app's model picker and in our list of service providers below.
Apps you connect, such as Slack
If you connect another app to Multilo — for example Slack — the desktop app signs in to it as you, with your own permissions, and reads or writes that app's content only when the assistant uses it for your request. That content is sent to the AI model you chose for that request and kept in your chat history on your computer; Multilo's servers do not store it. Your sign-in for the connected app is kept encrypted on your computer. The assistant posts or changes anything in a connected app only when you ask. Content from connected apps is never used to train AI models. To stop access, disconnect the app in Settings → Customize → Connectors, or revoke Multilo in that app's own settings; deleting a chat deletes the connected-app content in it.
The Multilo browser extension
The Multilo extension for Chrome, Edge and other Chromium browsers lets the desktop app work in your own browser when you ask it to. It talks only to the Multilo app on your computer, through your browser's native messaging, and sends nothing to Multilo's servers or anyone else.
What it can see: it works only in the tabs in its "Multilo" tab group — the ones it opens and the ones you share or drag there. It does not see your other tabs, your browsing history, your bookmarks or your saved passwords.
What it reads: when the assistant uses one of those tabs for your request, it reads that page's content (its text and structure, screenshots, console messages and the list of the page's network requests). That content is sent to the AI model you chose for that request and kept in your chat history on your computer; Multilo's servers do not store it.
Websites and sign-ins: Multilo asks you before it uses each new website. It never types passwords, payment details or other credentials for you; it hands sign-in pages and CAPTCHAs back to you.
Ask Multilo: when you use "Ask Multilo" or share a tab, the page's address, title and any text you selected go to a new chat in the app. Nothing is sent to a model until you press Enter.
What the extension keeps: only a list of the tabs in its group, in your browser's session storage, cleared when the browser closes. It has no analytics or tracking. Content from your browser is never used to train AI models.
Turning it off: use Settings → Agent → Your browser, close the tab or drag it out of the Multilo group to stop it there, or remove the extension from your browser.
Our use of information received through the extension complies with the Chrome Web Store User Data Policy, including the Limited Use requirements.