The problem JARVIS Bots solves

JARVIS is built with one singular purpose in mind i.e. maximize your coding time. It is a cloud-native Intelligent Process Manager (IPM) based on natural language processing and understanding that mines, analyses and serves information from the tools deployed from the corporation intranet. It uses the multiple information sources in order to learn about a product to optimize developer/product support processes in a team from task allotment to project deployment.

What differentiates us is:

1. We have taken the time to understand development workflows in various small and medium software development companies and all our use-cases are based around minimizing task accomplishment time for the same.
2. We also make it very easy for companies to customize as per the specific product development process they follow.
3. We also have the ability to deploy on a company's on-premises cloud. This becomes extremely crucial because most companies are extremely wary about their development data.

Challenges we ran into

We ran into 3 bugs mainly while working on this project that we did not have time to resolve for the project:

Accent: Since some people might have a heavy accent the text to speech model’s accuracy may suffer however we plan to combat this by retraining the default Amazon Lex Model on speech samples of Indian Accents.

Email Parsing: Despite the advances in NLP/NLU there might be case in where the bot might miss out on the context due to inadequacies in the algorithm’s capacity to hold the same over a long time especially, if the conversation is outside of a single thread.

Multi-word Intents: In certain use cases we miss out on picking up multi-word intent and the lack of a transcript makes it hard to apply custom NLP models for the same. This is especially true for creation-based use cases where multiple word data is commonplace.