The problem FileSphere.ai solves

People might not have the technical knowledge to manage file on their own. Or enterprise are facing the productivity issue in managing large database of files.

Let think about the perspective of an emplyee, the said "Read that file, summarize it, and convert it to pdf before senting to [email protected]" and employee like us might think what if there is an AI for this.

That why we build this. Just install it, open it and say it. The AI will do everything for you.

Example: Just say "Read that file, summarize it, and convert it to pdf before senting to [email protected]"
This AI software can identify the "that file" and it can execute multiple tasks sequentially. because it is build on top of LLM. So it can understand like human.

Managing files and executing various kinds of tasks in one prompt is very difficult. We are trying to make execute things with prompt. We built this in the top of the Large Language Model thats why basically it is called Large Action Model.

We actually build a core technology to convert normal LLM to smart LAM (Large Action Models). npm install initiative.

Challenges we ran into

File can't easy read through nodejs libraries, so we implemented golang to handle that part. Out custom core package has the type safety issue. Also it can't handle mulltiple extecution and storage of return values. So we implemented a in memory storage bucket.