Before I get comments saying I’m exaggerating, I want to note that there’s quite a lot of thing happening on the backend, its a pretty robust system and the examples are theoretically possible. My friend and I definitely lost some sleep and our minds trying to make this work.
WHAT IS IT?
We called this project Nelima and it’s a free AI community-driven tool that can execute any kind of tasks for you. It’s basically designed for taking actions on your behalf with natural language commands. You can integrate your own custom actions to suit your specific needs and you can layer multiple actions to perform more complex tasks. Users can actively influence the way the AI agent can work, cough cough with function calling. That’s why it’s “community-driven”, It's not just about what the tool does out of the box; it's about what it can do as the community grows and shares new actions. This collaborative effort can hopefully simplify setting up complex workflows.
An example I like to give would be creating a function/action that calls you an Uber every time the time hits 5:30pm from Monday to Friday. This action is then part of Nelima’s capabilities for other users. The "agent" gets to see the outcomes of each micro-action undertaken to achieve the overarching goal. As a result, the potential for reasoning is very much possible, and doesn't shy off of taking measures (lets say, it sees your grocery list from a sub-action on fulfilling an action and from it's own knowledge it realizes that a certain item has allergens which might be harmful for the user and puts a warning label, even tho the user didn't ask for this) to influence the direction of execution.
It’s actually a little bit similar to Zapier but instead of connecting apps with predefined triggers and actions or focusing on marketing tools, we’re trying to add an additional layer and add multiple “blocks” and interact with different software through an LLM. People just have to create those functions.
WHY BUILD THIS?
Right now, most language model systems focus on converting everyday language into more polished text or other. But there's a big opportunity in using these systems to turn everyday language into code that can control lots of different software programs. Imagine being able to tell your computer, "Put all my Tweets from last month into a Google Sheet and send the link to my Slack." I’m sure you’ve heard of this concept before but we’re trying a different community-driven approach.
IS IT THAT EASY?
Long answer: Heck no. Short answer: no. There are definitely some challenges. We’re working on building some adaptors so different systems can interact and actually process requests. Also, there’s so many things that could be done that a community is kinda needed to make this work (please build something, I beg). Some functions are easy to make, for example, we did one where you can ask the weather and Nelima goes and search online for it. Also working on getting it to read files/docs.
tl;dr there’s a lot of work ahead but I think it’s a super cool concept at least!
You can try it out here if you’d like: https://sellagen.com/nelima.
We're looking for people that join our discord to collaborate, troubleshoot, request and create some really awesome actions/functions/agents! Also any feedback on the project would be awesome!
Top comments (0)