Project: DocAssistant - AI powered app for answering questions for any documentation or documents

[YuriyMorozyuk95]

Project name

DocAssistant

Description

DocAssistant is the ultimate co-pilot for anyone who wants to breeze through documentation and get to the heart of what matters. Whether you're a customer support specialist or a budding business analyst, DocAssistant will be your trusty sidekick in navigating the world of documents.

With this cutting-edge app, you'll no longer need to spend endless hours poring over lengthy manuals or technical documentation. Simply upload your document, ask the questions that matter most to you, and let DocAssistant do the heavy lifting. It's as easy as that!

DocAssistant's intuitive interface makes it easy to scan and upload documents from a variety of sources. Whether you're working with a PDF file, an image, or even a repository from GitHub, DocAssistant is up to the task.

And for tech-savvy users who prefer to work with .md files, DocAssistant has got you covered. With the upload.git command, you can quickly and easily download an entire repository, DocAssistant will download all .md files from the repository, use it as a knowledge base to get the answers you need in a snap.

So if you're looking for a powerful tool to help you stay on top of your documentation game, look no further than DocAssistant. It's the ultimate co-pilot for anyone who wants to get the most out of their documentation without the headaches and hassles.

It all works thanks to the new Microsoft cognitive services Azure Open AI service, which was presented at Microsoft Build and for the fine-tuned of our model based on GPT-3.5 Turbo with with the generated greate context, system prompt and examples, parameters, for communicating with the model, which were varied and supplemented with code for each command. It was also necessary to use Azure Form Recognizer and Azure Bot, Azure App Service, App registration and other office 365 tools such as Teams Developer portal and so on in order to deploy the bot. The entire implementation of the project was done using .Net 7 in C# 11 and Microsoft.Bot.Builder 4.20.0.

I would also like to thank Alexander, our .net developer, business analyst and prompt designer Anastasia, and Margus, who was engaged in testing, for the all works that was done and sleepless nights.

Separately, I would like to thank Igor for the work he did in the video editing of this cool video, and Timofey, who helped us at the beginning of the project, unfortunately, due to lack of free time, he could not continue the hackathon with us.

Thank you for attention

Project Team:

Yurii Moroziuk – Project Lead /.Net Developer / DevOps
E-mail: [email protected]
Credly: https://www.credly.com/users/yurii-moroziuk/badges
GitHub: https://github.com/YuriyMorozyuk95

Alexander Perebaskin - .Net Developer
E-mail: [email protected]
Credly: https://www.credly.com/users/alexander-perebaskin/badges
GitHub: https://github.com/alx875

Margus Martsepp - QA
E-mail: [email protected]
Credly: https://www.credly.com/users/margus-martsepp/badges
GitHub: https://github.com/margusmartsepp

Anastasiia Riabova – Business Analyst / Prompt and AI engineer
E-mail: [email protected]
Credly: https://www.credly.com/users/anastasiia-riabova
Git Hub: https://github.com/Marzipanegg

Project References:

Project repository: https://github.com/YuriyMorozyuk95/DocAssistant
Project board: https://github.com/users/YuriyMorozyuk95/projects/2
GitHub Workflow action to deploy project: https://github.com/YuriyMorozyuk95/DocAssistant/actions/workflows/main_docassistant-api.yml

Project Repository URL

https://github.com/YuriyMorozyuk95/DocAssistant

Project video

https://youtu.be/FNM0-4_llHA

Team members

YuriyMorozyuk95, alx875, Marzipanegg, margusmartsepp