How AS Software Replaced 290+ Instances With One Platform

The customer is a market leader in the ultrasound segment, but its codebase was 20+ years old and not built for cloud or multitenancy, so every customer needed a separate deployment of the full stack. The economics only ran one way: each new customer meant another full stack to stand up and maintain, so the cost of serving a customer climbed with every deal. Running 290+ independent stacks made everything slower too, since releases had to be pushed to each instance one at a time and, with the frontend coupled to the core and nothing multi-tenant, features could not be built once and shipped everywhere. A bundled commercial image viewer added significant COGS, and rich imaging and study data went unmonetized. After a PE acquisition with a mandate to accelerate growth, the modernization had aggressive timelines that traditional development could not meet.

Ideas2IT ran Legacyleap, its Gen AI-powered legacy modernization platform, across the entire VB6 and .NET codebase before the rewrite, using its agents to read, document, and map the application's architecture, dependencies, and business logic and produce concrete modernization recommendations. That cut the overall rewrite by 60%. Ideas2IT then consolidated the 290+ single-tenant instances into a multi-tenant SaaS on AWS, decoupled the React frontend from the core, and migrated the full database to the cloud, so releases deploy automatically from a single source. The commercial image viewer was rebuilt as an AI-powered solution the company owns, removing an estimated $500K or more per year in per-customer licensing. Three production AI products shipped in the same 12-month window on Anthropic Claude with smaller task-specific models: AI Imaging Analytics, Voice-to-Text Documentation (cutting documentation time an estimated 50%), and Conversational BI.

Before the rewrite began, Ideas2IT ran Legacyleap, its Gen AI legacy-modernization platform, across the entire VB6 and .NET codebase. Its agents read the application in full, documented it, and mapped architecture, dependencies, and business logic, then produced concrete modernization recommendations. That gave the team a complete view of what lived inside 290+ instances without months of manual code archaeology, and it cut the overall rewrite by 60%. Ideas2IT then consolidated the 290+ single-tenant instances into a multi-tenant SaaS on AWS, decoupling the React frontend from the core so features could ship on their own, and migrating the full database to the cloud in the same move. Releases that once meant updating each instance by hand now deploy automatically from a single source. The commercial image viewer was rebuilt as an AI-powered product the company owns, removing an estimated $500K or more a year in per-customer licensing. With the platform rebuilt, three production AI products shipped in the same window: AI imaging analytics, voice-to-text clinical documentation, and conversational BI, all running on Anthropic Claude with smaller task-specific models where a lighter one suffices.

Software companies that ship a separate deployment per customer, where infrastructure or third-party licensing compounds with every deal and legacy architecture stands between the company and AI.