I'm going on two months of using Proton's Lumo as my primary AI and honestly, I'm not looking back.
Before this, Perplexity was my main AI because I found its answers to be particularly helpful on technical subjects. It sources things very well and tends to have the best overall answers with technical topics. However, it's a Big Tech AI service: it's inherently horrible for privacy. While my Perplexity account is pseudonymous and I only ever use it for things like troubleshooting and research, it's still not good for privacy.
Before I decided to give Lumo a proper try, I had been using Venice AI. It's a privacy-respecting AI service with end-to-end encryption options and uncensored options. At its core, it's a solid service itself, but the main problem with it is that it has severe restrictions on free users: only ten text queries allowed per day. That's an absurdly low cap. Worse still, their lowest plan is $18/month. I use AI a fair amount but not enough to justify an $18/month subscription plan.
When Lumo was first launched, I found it to be pretty useless. It hallucinated constantly, the UI/UX was barebones, it just wasn't a great experience. However, when Lumo 2.0 dropped, things changed. Now, it feels nearly on par with Perplexity in terms of information and sourcing, and the UI/UX is actually excellent now. I find it to be very rare that I ever have to use a different AI, even with their free plan restrictions, and I'm even to the point where I'm considering subscribing precisely because I don't feel like I'm exploited as a free user. Their subscription is ~$10/month, by the way: significantly cheaper than just about every other AI option out there, as even the cheapest alternatives are still heavily restricted.
The fact that an end-to-end encrypted, privacy-respecting option is now truly competitive is amazing news because I love the benefits of AI, I'm just not keen on the ways it can be used against us.