Launched this week
Lethe
See how you build, not just what you build.
9 followers
See how you build, not just what you build.
9 followers
The AI workspace that remembers your projects — not just your prompts. AI tools are great at solving problems, but they still make you restart every project from scratch. Lethe keeps track of your projects, goals, documents, and context so your work doesn't reset with every new conversation. Instead of remembering prompts, Lethe builds an understanding of your work that grows over time and can support multiple AI models. This is our first public release, and we're looking for honest feedback.
Free Options
Launch Team / Built With
Hi everyone! I'm Q, the solo founder behind Lethe.
I built Lethe because I kept switching between different AI tools and found myself repeating the same project context over and over. It felt like every conversation started from zero.
Instead of building another chatbot, I wanted to build a layer that understands the work I'm doing so projects can continue over time.
This is still an early product, and I'm launching to learn rather than to claim it's finished. If you try it, I'd really appreciate brutally honest feedback—especially about what confused you, what felt valuable, and what you'd improve.
Thanks for checking it out!
honestly the project memory concept is genuinely interesting, but i'd love to see a way to manually tag or label what lethe remembers, basically like folders or tags. right now i imagine it could get messy fast if i'm juggling multiple clients and the ai mixes up context between them. some kind of visual project switcher would make this way easier to trust day to day.
@fikret1238579 You called it — tagging/labeling projects is already in the works. Free-text tags with auto-generated colors (no manual color-picking needed) so multi-client juggling doesn't blur together. Should ship soon.
Congrats on the launch, the project memory layer is something I've wished existed for a while. One thing that would help me trust it more is a clear diff view showing exactly what context Lethe pulled into a response, so I can see why it answered the way it did and adjust the stored project info when it pulls in the wrong stuff.
@erkansulumutlu Really good point, not built yet. A "why did it answer this way" context view would go a long way for trust — I'm noting this as a real feature to scope, not just a nice-to-have.
One thing I'd love to see is the ability to tag or label projects so I can quickly filter context when switching between unrelated work, like separating client research from personal experiments. A simple color tag or short label on the project card would help me jump in without scrolling through a long list.
@hsantaknsoaalh Same answer as I gave Fikret above — project tags/labels are in progress right now, exactly the filtering-by-label idea you're describing. Appreciate the specific ask, it's shaping the design.
Congrats on the launch, the project memory idea is genuinely useful. One thing that would help me as a user: a way to export or snapshot a project's full context as a single file, so I could back it up locally or move it between machines. Right now I'd worry about getting locked into a single account if my data lives only in your system.
@bernahrlg Thanks so much! 🙌 That's an important point. Lethe is designed so you're never locked in. Each project's context can already be exported, and your structured memory is exportable as well. That said, your comment makes me realize I should make this much more discoverable—if people don't know the feature exists, that's on me. Thanks for pointing it out!
Finally tried this and the project memory actually sticks across chats, which is the main thing I've wanted from AI tools for months. Curious to see how it handles bigger codebases though.
@necatialtuntal Glad it's sticking for you! Bigger codebases are next on my mind too — curious to hear where it starts to strain once you push it further.
the project memory concept is really compelling, it always bugged me how every chat starts fresh like nothing existed before. love that it works across multiple models too
@adnan616769 Thanks! The "starts from zero every time" problem was exactly what drove me to build this, so glad that's landing. Multi-model support was a deliberate bet too.