What You Missed at the AI Agent & Copilot Summit 2026: TheTestMart was on the ground in San Diego. Here’s what stood out.
The AI Agent & Copilot Summit, produced by Dynamic Communities, wrapped last week at the Hilton La Jolla Torrey Pines in San Diego. Nearly 1,000 business and technology leaders gathered for three days of sessions on where AI is actually landing inside Microsoft Dynamics 365 environments. TTM was a proud sponsor, and we left with a clear point of view on what this moment means for D365 organizations.
IT is turning business needs into solutions
A consistent theme wasn’t about pure technology. It was about how teams are working together differently, faster. Kenny Mullican, CIO at Paragon Films, said it well: AI is giving technology teams the ability to support the business directly in ways they couldn’t before. Less time in ticket queues, more time solving real problems.
That showed up in the use cases shared. A major US manufacturer had been running traditional ML models to manage complex EDI workflows, but as those processes grew in complexity, they moved to AI, building D365 extensions mapped to specific APIs to pull exactly the data they needed. They’ve since applied that same pattern to fulfillment and transportation. AI isn’t replacing ERP. It’s making ERP more tailored to the specific needs of the business, and more mission-critical as a result.
Reskilling: outputs over prompts
A few things from the reskilling sessions stood out:
- Measure outputs, not prompts. How a team thinks with AI matters more than how often they use it.
- Make it safe to amplify AI use, not hide it. Better results with AI assistance is the outcome worth recognizing relayed, Shawn Dorward, Microsoft MVP and VP SMB Division at sa.global
- As Will Hawkins, CEO of RitewAI, put it: AI is linked to how you think. PhD-level thinking gets PhD-level results.
The natural language ceiling
Natural language is a starting point, not a control plane. That was the undercurrent across a lot of sessions.
Copilot and AI agents work well for surfacing insights and handling lower-stakes workflows. But for revenue recognition, inventory valuation, or compliance-sensitive transactions, “the model interpreted it correctly” isn’t something a CFO signs off on. D365 environments are often too complex and too customized to take AI-generated outputs at face value. Teams are building agents on top of systems that were never designed to be interrogated dynamically, and the complexity compounds.
Data: still the real conversation
If there was one topic that surfaced in nearly every hallway conversation, it was data quality. AI doesn’t fix bad data, it amplifies it. We had a great conversation with Conor Doyle, CEO of DQ Global, who works on exactly this problem. His framing was simple: before you can use your data to inform your deepest customer decisions, you need to sort out all the “Dan’s, dans, Daniels, Danny’s” living inside your system. The models are only as good as what’s underneath them, and most enterprise data isn’t ready for what’s being asked of it.
Takeaway
AI adoption in D365 is accelerating. More customization, more agents, more unstructured data flowing into structured processes. The builds are moving fast. The question every team should be asking is whether their validation strategy is keeping pace. Because the cost of finding out it isn’t keeps getting higher.
Is Your Testing Strategy Ready for What You’re Building?
AI adoption in D365 isn’t slowing down — and neither are the risks of untested outputs. Is your testing strategy ready for what you’re building? Book a free D365 Automation Assessment with TTM’s Head of Testing and get a clear, actionable picture of where deterministic, automated testing fits into your roadmap.
