AWS Transform SQL Server Modernization

AI-assisted cloud migration and modernization tool for full-stack SQL Server to Aurora PostgreSQL transformation

Timeline

June 2024 → GA Launch December 1, 2024

Role

Lead UX Designer

Scope

End-to-end database and application modernization workflow

Summary

I led UX for a 0→1 Agentic SQL Server modernization workflow within AWS Transform, launched at re:Invent 2025. The challenge was designing an efficient experience for enterprise customers to speed up their database migrations. As UX lead, I defined the workflow from assessment, wave planning to modernization execution in a chat driven experience.

View interactive demo

Problem

SQL Server to PostgreSQL modernization is a critical but daunting initiative for enterprises. Traditional approaches take months of manual work and require deep technical expertise across databases, applications, and infrastructure.

[Users and jobs to be done]

Challenge

How might we guide users through highly technical modernization workflows so they can act faster, with clarity and control—without being overwhelmed?

This challenge centers on two major shifts:

🧠 Designing for domain complexity and cognitive load: Managing technical complexity without overwhelming users with information.

💬 Moving from traditional console workflows to an agent-assisted, chat-centric experience: Transitioning from familiar UI patterns to conversational AI interactions.

Process

To address the complexity of designing highly technical, iterative workflows, I used rapid mockups, PM-eng-UX sync-ups, and technical SA reviews, then validated with 15+ customers. I partnered with platform designers to define key agentic UX patterns and prototyped new interaction models to validate usability with users.

[Design process and timeline]

💡 Key research insights

Users need guided, not fully automated, experiences: While users want efficiency, they don't want to lose control over critical decisions. This led to designing "human-in-the-loop" (HITL) interactions at key decision points.

Chat alone isn't enough: Users don't want to do everything in chat and expect it to be smarter. We balanced chat-based AI interactions with traditional UI for complex data visualization.

Final result

Designed an AI-assisted workflow that automatically analyzes SQL Server databases and .NET applications, converts schemas, migrates data, and refactors application code while maintaining human oversight at critical decision points.

[Chat-first, UI-supported paradigm]

Chat-first, UI-supported paradigm

Observation: Chat is powerful for guidance but poor for precision, bulk actions, and large datasets.

Decision: Designed a hybrid interaction system:

  • Chat for intent capture, explanations, and quick plan edits
  • UI for structured data, dependencies, bulk operations, and validation

This avoided forcing users into unnatural conversational inputs for technical work.

[Progressive disclosure]

Progressive disclosure

Layered information architecture from chat summaries to detailed data tables to drill-down views.

[Human-in-the-loop (HITL)]

Human-in-the-loop (HITL)

Clear decision points where users review and approve AI recommendations with full transparency.

Impact & outcomes

Adoption success: 117 customers adopted SQL Server modernization with 167 active users and 205 workspaces created since GA launch.

Efficiency gains: 911,676 lines of code analyzed and transformed across 11,674 servers with 90 completed jobs and 0 failed jobs.

Enterprise impact: Reduced modernization timeline from months to weeks, serving major customers including Accenture, Canadian Pacific Railway, and Commonwealth Bank of Australia.