dvloper.io at VMware Explore 2026, Las Vegas

In early September, part of the dvloper.io team was in Las Vegas for VMware Explore 2026, the annual Broadcom / VMware event for the people who build and operate private cloud. Four days at the Venetian, 400+ technical sessions, Hands-on Labs, certifications, and plenty of hallway conversations with people facing the same problems our customers face. Here is a short recap of what we did there and what we brought back.

What we did there

We worked on three fronts: technical sessions on Private AI and agents, the Private AI workshop, where we went hands-on with the VCF + Private AI Services stack, and direct conversations with Broadcom teams and ecosystem partners about what already works in production versus what is still on the roadmap. That last part matters: with AI, the gap between a slide and a shippable product is exactly what you need to understand before promising anything to a customer.

The technical themes that matter

The message of the year: AI is becoming a normal workload in the private cloud, and VMware Cloud Foundation 9.1 is the platform everything runs on (models, data, agents) without data ever leaving the datacenter. A few concrete things we took note of:

• Private AI Services in VCF: model serving, data indexing and RAG directly on existing infrastructure; model-as-a-service with validated models (Gemma, Qwen, Nemotron, GLM) selected per workload, rather than a single "official" model.
• VMware AI Factory: from hardware (GPUs, bare metal through the MetalSoft integration) to an operable AI environment, treated as an operating model rather than an isolated data science project.
• AgentMinder: the governance layer for AI agents, covering identity, authorization, policy and observability. Once an agent is calling APIs and touching data at machine speed, the question "who is it, what is it allowed to do, and what did it actually do" becomes mandatory. For us, the most relevant announcement at Explore.
• Tanzu Platform 10.4 agent foundations: agents become a normal application workload. AGENTS.md + manifest → cf push, service binding to a platform-hosted LLM and tools through an MCP gateway, with observability built in. No separate stack just because the application became agentic.
• AI-ready data: an ingestion, structuring, governance and lineage pipeline that publishes data products with semantic context for agents. The least glamorous topic, and the one that decides whether an enterprise AI project succeeds.
• AI in operations and security: AI Assistant for VCF (conversational diagnostics, AI-assisted Management Pack Builder) and AI assistants in vDefend / Avi for multilayer defense against AI-accelerated attacks.

This is exactly the direction we are already taking with our enterprise customers: private AI, on their own infrastructure, with agents that reach production only if they are governable.

dvloper.io on stage

Marian Tudosoiu, CTO of dvloper.io, delivered a presentation at Explore on The Database Modernization Ladder, explaining VMs to Containers on VMware Cloud Foundation.

What's next

Explore moves to Resorts World Las Vegas in May 2027. Until then, what we saw in Vegas goes straight into our projects. If you want to talk about what Private AI or AI agents would look like in your infrastructure, get in touch.