ai store

Thanks for checking this out! 🙌

We built this because we were frustrated with having the same conversation
with every AI app constantly re-explaining who we are and what we prefer.

Try it with the included Travel Planner, Packing Assistant, and Budget Planner — tell something to one, and another can use it (only if you approve).

This is early, and we need feedback, especially critical feedback:

1. Does the memory model make sense?
2. Does the My Context dashboard feel empowering or confusing?
3. Do you have any feature requests?

Your thoughts here genuinely shape what we build next.

Congrats on the launch!
A question: Does the store enable me to publish my agent?

This is a really impressive, Great work!💯

I like how AI Store introduces a shared memory layer that connects multiple AI apps, so users don’t need to repeat information every time. The permission-based access to existing context is a smart privacy-minded design choice, and the “My Context” dashboard where users can view, edit, and delete stored data gives a good sense of control.

That said, I noticed a few things that could make the product even better:

1. When I tried to test it myself, I couldn’t log in because I didn’t have any credentials. It would be great if users could get demo access or a temporary login to try the platform directly. This would help provide feedback based on actual interaction rather than only from the demo video.

2. In the demo, the agents used were pre-defined within the system. I was curious whether users will eventually be able to connect their own LLM accounts — for example, logging in with Gemini, ChatGPT, or Grok — so they can use their preferred AI models within the shared memory framework. If that’s not the case, then it would help to clarify whether the platform uses its own dedicated LLM for all interactions.

3. Regarding context management, I think it’s important to consider how long-term context growth will be handled. Even though some modern LLMs support very large context windows (up to a million tokens), research shows that performance often deteriorates as the input length increases, models tend to lose focus, overlook relevant information, or misinterpret earlier context. You can refer to this article for more information on the topic.

4. Finally, about the “My Context” interface, I think it could be simplified. Instead of multiple categorized sections with separate edit, delete buttons, you could just display the context file in a markdown viewer that users can view, edit inline, and update directly. It keeps things clean, fast, and easier to maintain.

Overall, it’s a brilliant foundation💯, I’m really impressed🫡 by the idea and looking forward to seeing how it evolves! 🚀

Hey @emmanuel_didymus_sebastian amazing idea and execution.
I think the way the context is carried across different agents is really cool and useful.
However, I found that my chats disappeared when I logged in on another device or when I logged our and back in.

The progress is great, I would only recommend a few tweaks on the UI side:
1. The send button in the agent chat area overflows in the mobile UI.
2. The visit site buttons also need to be responsive on the mobile UI.

3. Find a way to render the markdown in the chat box, so that the replies don't come out as raw markdown.
4. This is just me, but dark mode would be amazing 😂

I'll comment with more whenever I come across any that's noteworthy. Great work!