What happened
Proton, the privacy-focused productivity company, shipped Lumo 2.0 on Tuesday. Lumo is the public AI chatbot the company released last year, and this is its first big upgrade.
The headline additions are image recognition and image generation. You can upload a picture and have Lumo analyze or edit it. Like other chatbots, it can also generate imagery from a text prompt.
The second change is for Projects, the widget where you upload documents and work through Proton's other products, including email and cloud storage. Projects now come with persistent memory that the user controls. Lumo can recall your preferences across separate conversation sessions.
Proton also says Lumo 2.0 answers most queries up to 76% faster than the previous version. A new "thinking mode" is aimed at more complex problems and questions. Proton describes the release as re-engineered from the ground up.
Why it matters
Chatbot feature lists are starting to look like the same spreadsheet. Images, memory, a reasoning mode: Lumo is now ticking boxes that other assistants ticked a while ago.
What differs is the pitch. Proton sells privacy first, so memory is the interesting part. Persistent memory is useful and also the feature that makes people nervous. Making it user-controlled is the right instinct, though the controls themselves are what you should check.
The speed figure is a vendor claim with an "up to" attached. Treat it as a direction, not a benchmark, until someone outside Proton measures it.
This is a solid catch-up update for a privacy-minded audience, not a shake-up of the field.
What it means for Europe
Proton is a Swiss company, and Lumo is one of the few chatbots on the market from a European privacy brand rather than a US giant. For European users and teams who prefer their AI assistant to come from a company built around privacy, Lumo 2.0 makes the alternative more practical.
The tie-in with Proton's email and cloud storage is the real play. If your documents already live in Proton's tools, Projects lets the chatbot work with them without a detour through another vendor.
What to do with it
- Test the memory controls first.Before feeding Projects anything sensitive, find out what Lumo remembers and how you delete it.
- Run your own speed check.Compare Lumo 2.0 with your current assistant on your actual tasks, not on a slide claiming up to 76%.
- Try thinking mode on a hard question.Pick something multi-step and see whether it beats the default answer.
- Keep images boring at first.Start with non-sensitive pictures while you judge how well the editing and generation work.
Ciao, sapiens. The chatbot now remembers your preferences. Please make sure you still remember your password.


