Enterprise AI · Event Analysis
Five announcements from Chicago show a company claiming the runtime, not just the connector
Shashi Bellamkonda · May 13, 2026 · Chicago, IL
Enterprise data in motion today
Boomi customers globally
MCP-enabled tools in Boomi Connect
AI agents running on Boomi platform
Steve Lucas opened the Boomi World 2026 keynote in Chicago with an Isaac Newton reference. Objects in motion stay in motion. His number: only seven percent of enterprise data is in motion today. The rest sits in systems enterprises have paid for, protected, and never fully activated. That number is the most important thing said on stage this morning, and it is not in any press release.
This is my third Boomi World. The pattern across those events has been connectivity wins, partnership expansions, and product depth at the application layer. What happened today at the Hyatt Regency felt different. The announcements moved down the stack into runtime and infrastructure. That is a strategic shift worth naming directly.
Boomi Connect Is the Governed Gateway, Not One of Many
Boomi Connect, announced generally available today by Chief Product and Technology Officer Ed Macosky and Mani Gill, connects AI tools including Claude, Copilot, and Gemini to more than 1,000 enterprise applications through managed, Model Context Protocol-enabled tools. The Boomi AI Gateway provides built-in policy enforcement, cost controls, and observability on every agent call.
Worth noting from the stage: the product team credited a step change in large language model capability four months ago as the inflection point that accelerated Connect's development. That timing is significant. It confirms Boomi was responding to a real shift in what agents could do, not building ahead of demand.
The CIO question is direct. If your organization has Claude, Copilot, or Gemini licenses and your agents are reaching enterprise systems through ad hoc integrations or custom code, Boomi Connect is the governed alternative. The policy enforcement and observability are not optional features. They are the product.
"The one remaining moat that you have beyond your physical infrastructure is your data. Do not hemorrhage this to public models."
Steve Lucas, Chairman and CEO, Boomi — Boomi World 2026 Keynote
Lunar.dev: Boomi Claims Intelligent Prompt Routing
The announcement I wasn't expecting: Boomi announced intent to acquire Lunar.dev, founded by Eyal Solomon and Roy Gabbay out of Tel Aviv. Both founders joined Lucas on stage. The strategic case Lucas made was precise. As frontier model costs converge and open-weight models reach capable equilibrium with proprietary ones, enterprises will route prompts economically, not loyally.
Lucas said it plainly: enterprises always pursue the lowest economic option with the highest rate of return. Intelligent prompt routing is how that plays out in production. Boomi positions itself as the router that knows which model handles which task at what cost, based on complexity and data sensitivity, while keeping enterprise data from leaking to public inference endpoints.
The behavioral argument Lucas made on stage was sharper than most vendor narratives I've heard. He described the moment when an employee is about to paste proprietary data into a public AI tool. The hesitation. The mental calculation. That scenario plays out billions of times a day across enterprises. Lunar.dev, folded into Boomi's governance layer, is the technical answer to a human behavioral problem.
"Not Launching Boomi Claw"
Lucas said explicitly from the stage that Boomi is not launching "Boomi Claw" and is not joining what I've been tracking as the Clawconomy: the race by infrastructure vendors to claim the agentic AI stack through aggressive acquisition and vertical positioning. He said it to laughs. He meant it as differentiation.
The irony is that the Lunar.dev acquisition and the Red Hat collaboration are precisely the moves a Clawconomy participant would make.
The distinction Lucas is drawing is about architecture, not ambition. Boomi is not trying to own the model layer. It is claiming Context, Data, Orchestration, and now Runtime. That is a more disciplined position than most vendors in this space are articulating, and it is the one that survives when model economics commoditize, which they will.
Couchbase and the Agent Memory Problem Nobody Is Talking About
The Couchbase partnership addresses a gap that has been quiet in most vendor conversations: agents forget. Without persistent memory, every agent session starts from scratch. For enterprise workflows involving financial data, customer records, or multi-step approvals, that is not acceptable.
The partnership with Couchbase CEO BJ Schaknowski delivers persistent memory across sessions, millisecond retrieval at enterprise scale, and full auditability of every memory call through Boomi's Agent Control Tower. Every agent action is observable. Every retrieval is logged.
Schaknowski said it cleanly: enterprises need governance as much as intelligence. Couchbase provides the distributed, high-performance memory substrate. Boomi provides the control plane. Together they move agent memory from a demo capability to a production requirement.
Red Hat: Boomi Finally Claims the Runtime
The Red Hat collaboration is the announcement I find most strategically significant, and it connects to something I've been tracking across Boomi's product releases for two years.
Boomi is taking ownership of runtime and infrastructure in a way that iPaaS vendors historically avoided. The collaboration with Red Hat brings Red Hat AI and OpenShift together with Boomi Agentstudio, Boomi's intelligent model router, and the Agent Control Tower. The new Distributed Agent Runtime, announced today, deploys agents on-premises with locally hosted runtimes and language models, keeping sensitive data behind the firewall.
The architecture is the argument. Boomi's intelligent model router assigns agent prompts to the right model in real time based on task complexity and data sensitivity. Red Hat's Kubernetes-native runtime handles high-performance inferencing and integrated AI governance across hybrid cloud environments. Enterprises can run their own models, execute the Boomi runtime, and keep the whole stack where their data resides.
I've written before that Boomi's position at the connection layer between systems is difficult to displace once established. The Red Hat collaboration extends that position down into the execution layer. That is not a partnership announcement. That is a platform boundary expansion.
Boomi Orchestrate: The Demo That Told the Whole Story
The live demo by Patricia Moore was the most technically dense moment of the keynote. The healthcare revenue cycle use case: persistent EDI errors, invoices taking 45 days to process, fragmented execution across the revenue lifecycle. The conventional fixes are point-to-point integrations, manual coordination, or bots that still escalate to humans.
Boomi Orchestrate combined Boomi-built and third-party agents registered in the Agent Control Tower, built the solution architecture from a natural language prompt, surfaced an MCP registry inside the API control plane for adding tools from multiple providers including an Anthropic server, and delivered a business performance dashboard embeddable via the new embed kit without requiring users to open the Boomi platform.
Workspaces, a new account-level framework for organizing and isolating assets by project, appeared in the demo as an enterprise governance layer. Organizations building agents across multiple business units need exactly this kind of project isolation and access control.
The Customer Argument Is the Honest One
Post Consumer Brands CIO Balaji, speaking from the keynote stage, made the customer argument for integration without ever calling it integration. Five acquisitions, two more finalized in recent weeks, a model that buys brands rather than full companies, and constant pressure to move acquired system data into core platforms before business value evaporates. The integration layer is what makes acquisition economics work.
That is the unglamorous version of data activation. It is also the most honest one. Boomi's 30,000-customer base is full of organizations with this exact problem. The data activation narrative is not a rebrand for its own sake. It is the enterprise pitch for a constraint customers have been living with for years.
"When Steve asked why people still drive when self-driving cars exist, I answered: Trust. That word carried through everything that followed."
Shashi Bellamkonda, audience Q&A, Boomi World 2026
CIO/CTO Viability Question
Boomi is claiming runtime, not just connectivity. The Lunar.dev acquisition intent, Distributed Agent Runtime, and Red Hat collaboration are a coordinated move down the stack. Before you treat Boomi as integration middleware, decide whether you want it as your AI execution infrastructure. Those are different procurement decisions with different security, cost, and vendor lock-in implications.
Start by mapping which of your current agent workflows are sending data to public model endpoints that Boomi Connect could intercept and govern. That audit answers the question faster than any briefing.
Sources
Boomi. "Boomi Unveils Innovations That Power the Agentic Enterprise." Boomi Newsroom, 13 May 2026, boomi.com.
Boomi. "Boomi and Red Hat Collaborate on Production-Ready Agentic AI." Boomi Newsroom, 13 May 2026, boomi.com.
Boomi. "Boomi and ServiceNow Partner to Power Data Activation Across the Enterprise." Business Wire, 12 May 2026, businesswire.com.
Lucas, Steve. Boomi World 2026 Keynote Address. Hyatt Regency Chicago, 13 May 2026. Author's notes.
Moore, Patricia. Boomi Orchestrate Live Demo. Boomi World 2026 Keynote. 13 May 2026. Author's notes.
Boomi. "Boomi and Couchbase Partner To Power Enterprise AI Agents With Trusted Recollection, Connectivity, and Governance." Boomi Newsroom, 13 May 2026, boomi.com.
