AskMyDocs.ai

Imagine if you could cut through the clutter of your company's documentation, no matter what platform it's on, and find exactly what you need within seconds. That's what our service offers.

Comment

Demo

AskMyDocs.ai development environment username: [email protected] password: joe+demo1

please don't try to connect it with your own slack workspace as the app is not in the Slack App Directory yet, I have it under review right now

Inspiration

I had the issue of getting a ton of questions from our support team with most of the answers in our documentation. I would have to either dig through docs or have the support team do it, averaging 5-15 minutes for questions that we've already spent time answering (hence how we have docs on it).

What it does

So I built this tool to save myself time in answering questions. I call it “AskMyDocs dot AI”. It allows teams to query documentation, adding value to a resource the they’ve already invested in

How we built it

It's build using Svelte as the UI framework, FastAPI as the server, uses OpenAI for embeddings and queries, Pinecone as the vector database, Supabase (postgresql) as a database, and SendGrid for emails

Challenges we ran into

Learning how to use a vector database to segment customer data was a challenge but I was able to solve it with the help of the Pinecone documentation and reaching out to the support team.

Figuring out how to easily host it without messing with EC2 and other services was difficult. Eventually I found Railway and was able to host it very quickly, easily, and cheaply.

Accomplishments that we're proud of

Proud of taking this from a problem, to an idea, to a prototype as a jupyter notebook, to a locally hosted web app, to a fully functional SaaS product.

What we learned

Learned how to use a vector database, how to setup security and auth so only users can access the app, how to make an efficient database schema, how to build base classes for the server so that I can easily add new entity types, and how to create a user friendly experience.

What's next for AskMyDocs.ai

Next steps are finding beta users to provide feedback on the UX, the value, and what we can do to increase the value of AskMyDocs (by saving them time) to the point that it would be a no-brainer purchase for them.

In the future, I'd like to implement features to see which documents people are asking about the most so companies can see where to invest resources in updating documentation. It will become a tool for not just using documentation, but for figuring out where users struggle (based on their questions) so the company can improve their services..

Built With

Share this project:

Updates

Submission history