AI Estimator

Learning from the past to make the project feature easier. Automated estimations powered by neural insights.

Comment

Inspiration

In the contemporary technological environment, sustaining a project backlog can prove to be a daunting undertaking. It occurred to me that as we undertake numerous projects over time, we accumulate a substantial quantity of historical data. Why not utilize this data to enhance the future prospects of our project? The idea of utilizing previous accomplishments to enhance future endeavors inspired the creation of AI Estimator.

What it does

AI Estimator is not just another tool, it is a visionary approach to project management. Using previous project data, it develops a neural network capable of predicting issue field values with precision. AI Estimator offers predictions that can save time, reduce human errors, and bring consistency to backlog maintenance, instead of manually sifting through the project past.

How we built it

I kicked things off by setting up a robust backend with Forge, ensuring that all data interactions were secure and efficient. With the foundation laid out, I moved on to designing the frontend using React. Leveraging React's component-based structure, I was able to create modular sections of the UI that could be reused and updated easily.

With the basic structure in place, I incorporated Brain.js to train the neural network using historical project data. The choice of Brain.js was deliberate; it's a powerful library that offers GPU-accelerated neural network computations right in the browser, making it perfect for my application.

The integration of these technologies allowed the system to analyze patterns and predict issue field values, streamlining the process of backlog maintenance.

Challenges we ran into

Every project presents its own set of obstacles, and ours was no exception.

  • The understanding and optimization of the neural network was a significant obstacle encountered during the process of neural training. It was a delicate balance to make sure predictions were accurate without overfitting.
  • Preparing the historical project data for neural training was a challenging task in itself. In order to achieve success, the data needed to be normalized and in the right format.
  • The process of creating an interface that was both intuitive and functional took several iterations. It was important to ensure that the end-users, project managers, and team members were able to navigate and use the tool with ease.

Accomplishments that we're proud of

Our greatest accomplishment has been seeing AI Estimator come to life and perform with remarkable accuracy. Not only did we successfully integrate disparate technologies, but we also pioneered an approach that could redefine traditional project management paradigms.

What we learned

What this journey taught us was:

  • Neural networks have a lot of potential when they are used with relevant information.
  • The importance of user experience, especially when new concepts are introduced.
  • The ability to persevere and be flexible is crucial for transforming ideas into reality.

What's next for AI Estimator

The journey of AI Estimator has just begun. With continued innovation and feedback from our valued users, we think it will become an important part of the project management toolkit.

The horizon appears promising. We want to do the following things:

  • Expand the neural model to handle more Jira fields.
  • Introduce a neural network management system based on good practices.
  • Provide support and work hard to ensure that this application is bug-free and smooth in use.
  • Provide additional options for customizing the prediction model.
Share this project:

Updates