by DeeDee Walsh, on Jun 2, 2026 9:07:34 PM
Microsoft Build is always a signal event. It's where Satya Nadella and team show you where the platform is going and by implication, what gets left behind. This year's keynote in San Francisco was one of the most consequential in recent memory, and if you're responsible for enterprise applications still running on Visual Basic 6, legacy Access databases, or other aging stacks, there were five moments you shouldn't have missed.
I spent a career at Microsoft before working on modernization technology, including time on the Visual Basic team itself. I've watched a lot of Builds. This one had a different weight to it.
Here's what mattered and what it means for your roadmap.
1. The Enterprise AI Platform Just Got Defined and the Entry Fee Is Modern Code
Satya Nadella spent lots of time walking through what he called "the AI stack": compute fabric, models, context, tools, agent runtime, and security/governance. The center of gravity for all of it is Microsoft Foundry, which now hosts over 11,000 models and is being positioned as the full application platform for the agent era.
Foundry is a lot more than a model catalog. It's a runtime for long-running autonomous agents, a governance layer for enterprise compliance, a context engine connecting to your organizational data, and an optimization loop that makes your agents better the more they run.
The implicit prerequisite for all of it: your applications need to be cloud-connected, API-accessible, and running on modern data infrastructure.
VB6 applications, typically on-prem, COM-dependent, tightly coupled to Windows Forms or legacy database drivers have no on-ramp to this stack. They can't expose the APIs agents need to act on. They can't connect to Cosmos DB, Azure Search, or Fabric. They can't participate in the IQ layer that grounds AI responses in your real business context.
This is what we call the AI Readiness Tax: a compounding cost that accumulates every quarter you delay modernization. Microsoft Build 2026 just made it much more visible.
2. Your Competitors Are About to Build "Hill Climbing Machines". Can You?
The biggest strategic announcement at Build was Frontier Tuning.
Microsoft's message to enterprises was pointed: stop just consuming AI from a shared frontier model. Start owning AI trained on your private data, your workflows, your institutional knowledge. Build what Mustafa Suleyman called a "hill climbing machine": a continuous learning loop that compounds your competitive advantage over time.
The case studies were super interesting. A McKinsey-tuned MAI model outperformed GPT-5.5 on McKinsey-specific tasks at ten times lower cost. Land O'Lakes got over 90% accuracy on complex internal workflows that had previously stumped general-purpose models.
Now think about what's encoded in your legacy applications. Decades of business logic. Edge cases built up through hard operational experience. Pricing rules, exception handling, compliance workflows, all of it crystallized in code. That institutional knowledge is one of your most valuable assets.
But you can't train an AI on code it can't access. Running RL environments, generating traces, creating rubrics, all of this requires modern, cloud-accessible infrastructure. Organizations that modernize their applications first will be the ones who can turn their operational history into a proprietary AI advantage. Organizations that wait will be training on borrowed time.
3. The Agent Economy Runs on APIs Your VB6 App Doesn't Have
Every demo at Build showed agents doing the same basic thing: connecting to real-time data sources, calling tools through standardized endpoints, reasoning over organizational context, and taking action across cloud-hosted systems.
The underlying architecture requirement is consistent: REST APIs, modern data layers, cloud-hosted execution environments. Microsoft's IQ layer: WebIQ for web grounding, FabricIQ for operational data, WorkIQ for procedures all assumes applications that can surface their data through modern interfaces.
VB6 applications weren't designed for any of this. They were designed for a world where the application was the system and not a participant in a larger orchestration layer. COM interfaces, MSAccess backends, and on-prem deployments are architecturally incompatible with how agents are built to operate in 2026.
This isn't a criticism of the engineers who built those applications. It's a mismatch of eras. But how long can your organization afford to maintain that mismatch while the operational gap widens.
4. AI Coding Tools Are Accelerating But They're Optimized for Modern Code
Microsoft announced MAI Code 1 Flash: a 5-billion-parameter coding model that achieves 51% on the SWE-bench Pro benchmark, optimized specifically for VS Code and GitHub Copilot. It joins a growing roster of code-native AI models that are all trained on modern languages, modern frameworks, and modern patterns.
This is a double-edged development for legacy shops.
On one hand, it means AI modernization is becoming capable and cost-effective. The models that power tools like VELO are getting faster and cheaper, which is good news for organizations looking to modernize at fixed cost.
On the other hand, it means the productivity gap between teams working in modern stacks and teams maintaining legacy code is widening really fast. Developers working in .NET, TypeScript, or Python have access to AI pair programmers, automated refactoring tools, and multi-agent review pipelines. Developers maintaining VB6 codebases are largely working alone, without tooling support, in an environment that the entire AI toolchain has moved past.
Every month you delay modernization, that gap grows.
5. The Microsoft Seller Motion Just Shifted And Here's What That Means for Your Budget
This one is practical and immediate.
Microsoft's entire go-to-market is now aligned around Microsoft Foundry and the agent platform. Every Microsoft seller in your account is being equipped and incentivized to help you get onto this stack. Cloud migrations, AI workloads, Foundry deployments; all of these are the conversations Microsoft wants to have with your CIO.
That creates a budget dynamic worth understanding. Enterprises that can frame application modernization as an Azure migration and AI enablement project are speaking directly to the story Microsoft sellers are trying to tell. Co-sell investments, migration credits, partner funding; all of it flows toward the Foundry story.
Organizations running on VB6 who don't have a modernization roadmap are going to find themselves in a difficult position in those conversations. The "we'll get to it" answer gets harder to sustain when Microsoft's entire field organization is asking about your AI readiness.
If you have a Microsoft relationship, Build 2026 just gave you new language for a conversation you may have been deferring.
What to Do With This
If you're responsible for applications running on legacy stacks, Microsoft Build 2026 is a useful forcing function. To be clear, legacy applications don't stop working on June 3rd. But the compounding cost of staying put just got a clearer price tag attached to it.
The good news: modernization doesn't have to be a multi-year death march anymore. Automated approaches like VELO can deliver a fully modernized application at a fixed price, without the open-ended risk that has made legacy modernization projects historically painful.
The VELO Assessment takes about 15 minutes and shows you exactly what percentage of your application can be modernized automatically with no commitment required. It's a good first step for any organization trying to understand what modernization actually looks like before committing to a path.
GAPVelocity AI is the AI modernization business unit of Growth Acceleration Partners (GAP), with roots in the original ArtinSoft technology that powered Microsoft's Visual Basic Upgrade Wizard. We've been working on this problem longer than almost anyone.
