grouchy

ofc!

I agree this is a real problem.

We will need a new set of tools to help developers with this.

Just like a whole set of web analytics, logging, feedback tools were built when we moved to the web.

Before you could just gather feedback from your employee interactions, or by watching customers in your store.

I don't see this as categorically different, but the tools do not exist today.

Why do you need *external* documentation but for helping people to understand how software works?

These apps will need lots of documentation but instead of having to "search" for the section you need it would be just exposed as you need it.

"How do I do X?"

You do X by filling out this form: [FORM]

I'm not sure the problem here?

Do you mean users or developers kowning the right questions up front?

Yeah, exactly. This is the progressive disclosure.

AI can give you a smart starting point (like a pre-filled formula) instead of making you start from scratch, and then you dig deeper only if you need to. That's what I mean by personalized software.

You get something that actually works right away, and you can tweak it from there. No need to watch YouTube tutorials or read docs just to figure out where to start.

I totally agree with this, and I'd go even further.

When I'm having trouble with software, I often turn to Google to figure out how to use it. I'm then directed to a YouTube video, help article, or blog post with instructions.

My take is that people are already accustomed to this question-and-answer model. They're just not used to finding it within the application itself.

I think the problem is having to "learn" software in the first place. You don't have to "learn" how to work with a good accountant or lawyer. They make it easy by exposing what they can offer precisely when you need it.

That's how I think software will work in the future. I'm not suggesting that the UI should be completely different on every render. Some predictability is essential. That's one reason I don't think codegen on every render is worthwhile. I'm simply suggesting that software should look different from user to user based on their individual needs.

my wife wants to learn guitar. i'll get her to use this and get back to you.

how did you build this? what tech stack?

Looks like you are using tambo-ai. Thanks for trying us. Let me know if you have any feedback.

Thanks for using tambo! Let me know if you have any feedback on how we could improve it!

Ofc needs tool calling and graphical UI. This isn't 1997.

Hi, thanks for the feedback. Good luck with Coldran.

To answer this question:

> why would anyone use this instead of building their own unique ui

0. Under our component library is tambo-ai/react, which are just React hooks, so you could build your UI from scratch (we had a user do this over a weekend). They apologized, "sorry I didn't use your ui components".

1. Most of our components are just for managing sending messages, showing chat state, etc. The rendered or intearcted with components are up to the developer and often pulled from their existing app (chart, forms, tables, etc.) and styling is up to them.

2. All of those provided UI components are the “copy and paste” model (like shadcn). You can easily update their styles to match your UI or customize them by simply changing the tambo.css file we add to your project.

> Internal Dashboard Chatbots can allow people to interact with some api endpoints which other components use within the website with a prompt and then the ui appears ~ that would be cool like creation of projects or changing the git connection from one repo to another without changing the pages.

We have this experience inside our dashboard https://tambo.co/dashboard you should try it!

Thanks! We often hear that people can spin up a demo of an AI assistant with MCP and their existing components in a few hours.

One person gave a demo to the entire eng team after a weekend!

We have a very similar model! The biggest differentiator from the NextJS/AISDK chatbot is that we have better support for interactive React components.

If you visit our landing page, you can see a demo: https://tambo.co

Our demo chat has an Email component registered clientside that the assistant can render; it also utilizes tamboState hooks to pass the state back into the chat history.

There are a few other differences, but that is one of the bigger ones.

I agree with the screen real estate thing.

One example I often give is a salesperson out on the road. They could talk to their CRM app:

"Add note to Joe we talked about his son going to college...also he's interested in getting a new tractor...he mentioned his buddy stan is looking at tractors."

The app will generate UI components inline:

Add Note (with generated note to approve/edit).

Add Opportunity (with SKUs filtered to tractors, but needs the user to select which tractor--since the user didn't say)

New Contact (with name generated and linked to Joe, but has some ui elements left blank for the user to enter, contact, company name etc.)

We are already seeing this play out in the developer tool space with OpenAI Codex, Claude Code github actions, etc.

Thanks! We have spent a considerable amount of time on the documents. It's the front-end of a developer product!

There is so much more to do on the front of managing composable and disposable (I use the term 'ephemeral,' but maybe that's too highbrow).

hey i gave this a test run. not sure it worked how I expected I thought it would index the entire site?

is that an offering of the product?

A ui is worth 1,000 tokens (at least).

I agree that chat isn't the end interface for AI powered applications.

That said the ability to describe desires in "fuzzy" ways is a usability value unlock.

My take is that natural language should be use surface functionality, translate between structures, and generate personalize defaults.

Its a hybrid approach and why my co-founder and I started https://tambo.co

hydra-ai is a react package that powers adaptive UIs with AI.

It’s open-source and free to start using.

hydra-ai 0.0.38 now supports:

- openai (o1-mini) - anthropicai (sonnet 3.5) - gemini (gemini 1.5 pro) - mistralai - grok - openrouterai - bedrock

Thanks to tokenjs!

I'm curious: How do you know it's accurate?

This is a really interesting use of LLMs for filtering content semantically. I feel like this has more general applications than just youtube/screentime filtering.