Re/Intro Feature Reclaimation

Challenge:
Recover valuable feature ideas that were lost during development.

Solution:
Create a set of AI agents to recover features, audit them for ROI potential, and summarize them for executive presentation.

Features

  • Smart Forensics For Agile Development

  • Feature Extraction

  • Feature Ranking

  • Data Visualization

AI Models

  • RAG Pipeline

  • Gradient-Boosted Trees Model

  • Large Language Model

User Experience

  • SharePoint Dashboard

  • SharePoint List View

  • Journey Map

  • ML Pipeline

Sharepoint Templates

Challenge

3rd Party Dashboard
A threaded data visualization to facilitate understanding of potential and scale.

 

3rd Party Matrix
A presentation of distinct assets for better pipeline management.

User Journey

A cohesive look at the entire product experience.

The purpose of the journey map is to expose a universal narrative of the Product Experience. It shows how a user moves through the product stages while at the same time showing where technology might intersect.

A user is able to see how the product works, and where AI will is engaged.

Agentic AI

  • The Historian

    Retrieval Augmented Generation
    (e5-large-v2 + SaaS GPT4)

    The Historian is an AI agent with the ability to extract features from Jira exports. Then it can identify which features are based on semantic understanding.

    It is comprised of two ML models, with one containing a neural network. Each model has to be benchmarked before joining.

  • The Auditor

    Gradient-Boosted Trees Model
    (XGBoost1)

    The Auditor is capable of researching various features and providing a more cohesive explanation for the ROI of each feature.

  • The Conductor

    Large Language Model
    (SaaS GPT4)

    The Conductor will review the highest-ranking features and provide a summary and a detailed specification to others.

University of Pennsylvania
Artificial Intelligence for Business

During this program, my team was tasked with creating a SaaS AI tool that could be used to automate product development and roadmaps. This design was based on common gaps inside agile-driven companies.

There was a need to create a way for business executives to be aware of the valued features that were relevant to past versions.

Remembering what Matters

In the past two decades of working inside of Agile development environments, I saw how features identified by stakeholders represented by Sales, Customer Support and Professional Services can often be lost in the bustle of scrum.

When assigned the task of involving AI in present development practices, I couldn’t help to think that what an executive team needs is a way to stay aware of precient features.

To acquire features based on concepts the customer base valued was a key goal of this product.

AI Agents to the Rescue

Beyond the fanfare of agentic AI, it was important for executives and non-technical stakeholders to identify with agents at a collaborative level. AI agents are built to help, and that matters most. As these agents evolve, understanding and support of their journey evolves as well.

Vignette involves a portalan nautical compass rose rendered by Ideogram.

 

Features

  • Smart Forensics For Agile Development

  • Feature Extraction

  • Feature Ranking

  • Data Visualization

AI Models

  • RAG Pipeline

    • e5-large-v2

    • SaaS GPT4

  • Gradient-Boosted Trees Model

  • Large Language Model