Capture what you know. Share a link. Visitors ask, your AI answers from your own sources.
No prompts. No agents. Drop things in, point a lens at a slice, share the URL. The whole product fits in three nouns.
Type, paste a URL, upload a PDF, record a voice note. Mindola transcribes, OCRs, chunks, indexes.
Group sources into Spaces. Each Space is a slice a lens can read from. Same brain, infinite faces.
Pick the Spaces it can see, set a tone, hand out the URL. It answers from what you gave it, and refuses cleanly when it can't.
Mindola doesn't care what shape your knowledge is in. Upload it, point a lens at it, share the URL. The lens figures out what to say from what you gave it.
A slice of your knowledge with a URL. Replace any noun and you've got a new lens. Six illustrative setups below.
When the lens answers, you're never out of the loop. Take over when something matters. The visitor sees "From Ana", not a bot.
Record sixty seconds. Mindola clones your voice with ElevenLabs and bolts it onto a lens. Visitors can call.
Read a short script with your name in it. After that, every answer your lens gives can be heard, not just read.
Every new account starts on a Pro-equivalent trial, so you can publish a lens before you decide. No card up front. Cancel and your lens unpublishes the same day.
Capture and chat with your own brain. No public lens yet.
A real lens, with a custom URL and an inbox.
Ten lenses, voice clone, Slack, and a custom domain.
* Custom domain is rolling out gradually - request enablement after signup.
The whole point of Mindola is that one person's knowledge has leverage. That only works if the person trusts the bucket. Here are the rules we hold to.
We chunk your sources into a vector index that's hard-bound to the lens. Out-of-scope questions are refused, not improvised.
Sources, chunks, embeddings, voice clones, conversation logs. One click, irreversibly, across every backup window.
JSON archive of everything you've captured and every conversation a lens has had. GDPR Article 20, no email required.
Your captures are not used to train any model, ours or anyone else's. Voice clones are scoped to a single lens.