AI Developer Assistant

a journey map showing personas on the left and flow on the right, with little stars scattered throughout
DXA Journey with key highlights in the workflow and identification of personas for which the AI Assistant would be most beneficial.
a scorecard that shows six categories of rating the generation output of AI responses
Scorecard
a list of scores assessed per question in an excel spreadsheet and the quality of each one assessed
Preliminary manual scoring for handoff to Developers
screen shot of UI designs in figma
Screen Mockups for UI *small to protect confidential
an experience map template for the development flow throughout the various stages
Experience Map for future AI integrations for entire flow

Impact & Results

Led efforts to leverage Generative AI by creating a RAG database, supporting development team processes and optimizing the software development lifecycle in order to improve the Development Experience Assessment score to at least a 75% from a 59%.

Problem Statement

Our engineering teams face significant challenges with the internal development process due to dense documentation, fragmented discussions, and a trial-and-error approach, leading to delays and inefficiencies. Teams spend up to four days on initial setup before actual development begins, and the reliance on inefficient self-support methods detracts from innovation, limiting the organization’s growth. To address this, we aimed to streamline the process with intuitive guidance, customized assistance, and reduced setup times, allowing teams to focus on innovation and core activities, ultimately enhancing productivity and accelerating development velocity.

Process & Approach

Following the Developer Experience Assessment (DXA), it became clear that an interactive method for seeking help was essential for both setting up repositories and managing advanced features like actions and runners. By applying UX practices such as personas and utilizing research from the DXA, I developed over 500 targeted questions to address our audience’s needs. Additionally, through research and collaboration with the Responsible AI team, I created a comprehensive scorecard. This scorecard enabled the development team to automate the process, ensuring that our AI assistant delivers useful and relevant results.

Tools & Methods

PowerPoint
Excel
Yammer
ChatGPT
Figma 
Miro
Jira | GitHub Loop

Leveraging conducted DXA research
Created scorecard based off industry standards
Extracted conversation data from yammer for question creation
Extracted insights from 13,000 pages of support tickets using assistance from ChatGPT to create scripts and internal chat tool to extract insights (protect privacy) for question creation
Designed mockups in Figma
Miro board of Experience Map
Work tracked in Jira & GitHub
Product Documentation tracked in Loop