Building AI-Powered Workflows into Enterprise HCM Software

Key Results

Challenge

Our customer wanted to explore how AI could become part of the software experience, not simply an additional tool alongside it.

That meant creating an AI experience that could support more than conversation. Users needed ways to interact with AI-generated information within the application, while the platform also needed capabilities for managing conversations, removing personally identifiable information (PII), testing new functionality, and controlling releases.

At the same time, product and engineering teams were exploring how AI agents could support the work of building and delivering software. 

Building the Command Center

Mindex created Command Center, an AI canvas that brings together AI Chat, interactive widgets, and file storage.

The platform uses generative UI, which allows AI-generated responses to be dynamically displayed within the application's interface. Widgets can have different behaviors based on the type of interaction, allowing users to work with AI-generated information directly within the application.

The platform also includes thread history and naming, along with capabilities to detect and remove personally identifiable information (PII) from conversations.

Testing workflows and feature switches were incorporated into the development process. Feature switches allow new functionality to be controlled during development and release.

Using AI agents across the development lifecycle

The work extended beyond the product experience.

Mindex introduced an agent-orchestrated software lifecycle in which human teams define goals, constraints, and desired outcomes, while AI agents support execution-heavy delivery steps.

The delivery model is supported by platform investments and specialized pod workstreams spanning product, engineering, and operational outcomes.

These workstreams include intelligent orchestration and labor budgeting, Gemini and Slack, outcome-based billing, approvals, personalized experiences and feedback, analytics, AI-powered setup, and Time-Off and Accruals Guided Setup.

The model is designed to support faster iteration, better scale, and more focus on high-value work.

Evolving the way product teams work

The project also introduced changes to the Product Management operating model, with an emphasis on value creation over process ceremony.

Rather than breaking work into highly prescriptive tasks, teams define work at the feature level, with engineering teams responsible for the detailed implementation work.

This gives Product Managers more time to focus on market analysis, customer engagement, commercial opportunities, product discovery, and long-range roadmap direction.

It also strengthens alignment across business, marketing, UX, operations, and engineering.

The impact

Together, our teams established an AI-enabled platform that combines conversational AI, interactive widgets, generative UI, workflow capabilities, and controls for managing AI conversations and software releases.

The work also established a framework for using AI agents within the software development lifecycle, with people defining the goals and AI supporting execution.

The result is a foundation for continuing to develop AI-powered experiences within the software and across the teams building it.

Meet the Team Behind the Work

A special thanks to the Mindex LBT and LBT Acceleration teams for their engineering expertise and collaboration throughout this initiative. Their work helped bring AI-powered capabilities into the software experience and accelerate development through modern engineering practices.

Ready to explore what AI-assisted development could do for your team?