Why are we shutting down?

tl;dr: We’re shutting down Tambo Cloud (Tambo stays open source and self-hostable) and pivoting to Charming, a collaborative cloud for apps built with any AI agent.

Why are we shutting down Tambo?

I wanted to start this post by thanking everyone who used and shared Tambo. We wouldn’t be here without you. It’s made the decision to explore other options, and finally shut it down, quite hard. It’s hard not to feel like I let you down.

It’s crazy to think that Tambo started out as a hackathon project idea. I was building an app that used an agent to generate UI. At the time, AI models could literally not generate code reliably. Fast forward to today, and they can generate front-end code far better than me or anyone I know. I thought this would happen, but not on such a short timeframe. Never underestimate model training progress. I’ve talked with many of our users, and they are also seeing changes that make what we were building with Tambo no longer necessary.

The truth is we experienced a lot of interest in Tambo, 11k GitHub stars, and thousands of people tried it, but we didn’t see the adoption that would have justified continued investment. And trust me, we invested a lot in Tambo, even back when you still had to write code by hand.

We made the hard decision to start exploring other options six months ago, and if I’m honest, I’ve been putting off shutting down Tambo for far too long, but it’s hard to throw out your babies.

What does this mean for Tambo users?

We have already reached out to everyone actively using the cloud platform, let them know about this shutdown, and offered to help them transition to a new platform. Tambo Cloud keeps running until October 31, 2026, after which the hosted API stops; user data gets deleted on November 30, 2026, and new signups are closed.

Self-hosted users can continue to use Tambo, but will need to maintain it themselves (the self-hosting docs are here). We are of course happy to help anyone with any issues they can’t fix themselves.

What are we building now?

The team remains mostly the same; we did part ways with a few people (amicably, I might add). The current team is rocking and rolling on the new direction. I wouldn’t be here without them.

Our new product, Charming, a collaborative cloud for personal apps built with agents, is still quite early, but we already have some very happy and active users. As a startup founder, you want to go where you feel pull, and we are feeling far more pull towards this direction than we were towards Tambo.

How did we find this new direction?

Well, a lot of it came from our Tambo users. Far more of them than I expected were only using Tambo to build apps for themselves, or that’s what they wanted it for. Actually, many of our early users of Charming came from those users!

Secondly, we were building more personal apps ourselves, sometimes because we wanted something unique, but often because we wanted to work with our Claude, ChatGPT, or openclaw/hermes.

Finally, we’ve been seeing the trend that more and more people don’t want to use some random AI assistant. You want to use the AI that you use every day, that has your context, that can work on tasks across products. With Tambo we were no longer building where the puck was going, but where the puck was.

Enter Charming. In truth, Charming was built for me. I was Charming’s first user, and the most active user at first. I personally hate setting up cloud hosting, auth, databases, getting OAuth tokens, etc., but I wanted to build apps for my friends and family. Charming abstracted away all the parts of building I didn’t like, and left only the parts I did. The parts that felt personal, that I was excited to use, and excited to show people.

The first “nugget” of pull I felt was when a school teacher I had never met in France became more active than me, including my personal and testing usage. Now I’m nowhere near the top users, if you don’t count the viral apps I shared on Reddit (kind of cheating, really). While we still don’t have tons of users, the ones we do keep coming back, most weekly, and a surprising amount daily. We don’t have PMF yet, but we feel much more confident we are on that path.

What’s next?

Well, who knows, but the team plans to keep working on Charming tooth and nail. We believe even more in a world where everyone can build the perfect app for themselves.

— Michael