Inspiration

11ku7-ai-chat is a simple ai chat assistant that work in your browser, it draws inspiration from various desktop ai chat interfaces, e.g. ChatGPT, Claude desktop etc. 11ku7-ai-chat has minimum and precise ui elements for interacting with llm. It has beautiful user interface with dynamically adapting for both mobile as well as desktop site. Most important use of this chat assistant is to access local models through ollama provider without any hassles, if we setup this ai chat assistant locally it will auto fetch model list through ollama server and has very short setup time. there is one more important functionality of this chat assistant is that it uses unique pipeline to handle attached pdf documents- uploaded document with multiple pages is split into separate document pages, and then those separate pages are converted into images, and then those images are queried with vision capabilities of llm.

What it does

it has three modes-

  • *normal mode: * it is the basic mode to chat with ai, this mode has various functionalities too like we can provide custom system prompt to change the behavior of llm, also the attachment functionality supports all type of images and documents to be attached in order to query them through vision capabilities of llm.
  • *code mode: * when this mode is enabled response will be precise code according to query without any explanation in llm response.
  • *webapp mode: * it is my favorite mode to create web application as a single html file containing html, tailwind css and javascript code for the web application as per given query by user.

How we built it

I built 11ku7-ai-chat in 2 days with the help of bolt.new and also with my own manual troubleshooting capabilities, overall, it was fun to use while it was building through each iteration through bolt.new. i dont think it will be possible to create this web application so fast other than bolt.new.

Challenges we ran into

There were so many errors while iteration on this chat assistant built up, but bolt.new handled errors very impressively. the most important challenge of this project was to enable attachment functionality in order to attach images and documents and handle them correctly in this ai chat assistant.

Accomplishments that we're proud of

  • 11ku7-ai-chat can be used both with hosted as well with local version.
  • it has modern ai chat assistant functionality in a minimum element-based user interface.
  • it can be used on both mobile devices and desktop.
  • it supports local model through ollama

What we learned

this is first version called as 1.0.0 of the 11ku7-ai-chat there we plenty of learnings-

  • creating user interface with modern framework like vite was very impressive.
  • implementing auto fetch model list for provider like gemini, openai and ollama make this ai chat assistant hassle free in setting up quickly.
  • all attachment capabilities support make this chat assistant perform similar to modern chat interface like chatgpt etc.

What's next for 11ku7-ai-chat

next version of the 11ku7-ai-chat will be 1.0.1 and there will be new modes added to it like mcp support, which will make this ai chat assistant more powerful.

Built With

Share this project:

Updates