Every CMO has seen the chart by now. Organic search traffic down, AI Overviews up, and a growing share of buyers never clicking through to your site at all. The conversation about AI search visibility has mostly focused on defense: how do you get cited, how do you get summarized correctly, how do you avoid being left out of the answer.
There's a second front opening up, and it's offense. It's called Agent Skills, and B2B vendors are already shipping them.
Boomi Just Showed What This Looks Like
At Boomi World 2026, Boomi released Boomi Companion, an open source suite of Agent Skills built on the Agent Skills standard that Anthropic and others have been developing. Skills are packaged expertise. Reference material, prompts, and tools bundled together so an AI agent, whether that's Claude Code, ChatGPT, or another assistant, can act with deep, accurate knowledge of a specific platform instead of guessing from general training data.
Agent Skills sit on top of open protocols like the Model Context Protocol (MCP), which Anthropic introduced to let AI agents connect to tools and data sources. MCP is the connection layer. A skill is the expertise layer built on top of it, the part that teaches an agent how to actually use a product correctly once it's connected. For marketing leaders, the distinction matters less than the outcome: agents can now be handed structured, vendor-supplied knowledge about how a product works.
Boomi also announced Boomi Connect, a managed Model Context Protocol (MCP) connector service that lets agents work with Boomi systems under governance controls. Together, the two moves point at the same idea: don't wait for AI models to learn about your product secondhand. Hand them the manual directly.
The integration skill alone reportedly ships with close to 25,000 lines of validated reference material. That's not a marketing one-pager. That's documentation depth aimed squarely at how AI agents now research and recommend tools.
Why This Matters for Marketing, Not Just Developer Relations
It's easy to read this as a developer experience story and move on. That would be a mistake.
Think about how a prospect now evaluates a category. They open an AI agent and ask it to build something, configure something, or compare options. The agent reaches for whatever knowledge is available, either baked into its training or available through a connected skill. If a buyer asks an agent to build an integration workflow and one vendor in the comparison has published a skill with real configuration patterns and error handling while the other hasn't, the agent doesn't just describe the first vendor's product better. It can actually build with it. The second vendor doesn't lose the deal because their product is worse. They lose it because the agent didn't know how to use it.
That's the shift. AI agents don't just recommend what they know about. They recommend, and increasingly act on, what they know how to execute.
This is the AI search visibility problem, one layer deeper. It's not just about whether your content gets cited in an answer. It's about whether your product gets chosen and used when someone asks an AI agent to actually do something.
The Skill Is the New Knowledge Base
For years, B2B marketing teams have built knowledge bases, help centers, and developer docs primarily for human readers, with SEO as a secondary consideration. Agent Skills flip that. The primary reader is now an AI agent acting on behalf of a user, and the skill needs to be structured so that agent can reliably do correct things with your product.
That changes what "good documentation" means. A skill needs to cover the patterns an agent will actually be asked to execute: how to configure something, how to troubleshoot common errors, how your product's terminology maps to generic concepts, and where the boundaries and guardrails are. Boomi's approach, building the skill around real platform knowledge, including known error patterns and best practices, is the model worth studying, regardless of your category.
Note that Boomi's figures on lines of reference material and the "35-plus compatible AI tools" claim come from Boomi's own announcements and haven't been independently audited. Treat them as directional, not verified benchmarks.
The Open Standard Is the Opportunity
The part that should get marketing leaders' attention is that the Agent Skills format is open. It isn't locked to one vendor's ecosystem. A skill built for one platform can, in principle, work across multiple AI tools. That lowers the barrier significantly. You don't need to negotiate a partnership with every AI vendor to get your product represented well. You need to publish a skill, the same way you'd publish a developer doc or an API reference, except this one is written for an audience of agents.
For B2B software vendors, this is a content asset that didn't exist eighteen months ago and is now becoming table stakes in technical categories. The companies moving first are setting the pattern that AI agents will default to when users in their category ask for help.
What to Do Monday
Get your product and engineering teams talking about whether an Agent Skill makes sense for your platform. Specifically:
Identify the tasks customers and prospects most often need help with, the ones that currently live in support tickets, onboarding calls, and developer forum threads. Those are the candidates for a skill's reference material.
Audit what AI agents already "know" about your product today. Ask Claude or another assistant to explain how your product works, configure a common use case, or compare it to a competitor. If the answers are vague, outdated, or wrong, that's the gap a skill closes.
Pull in whoever owns developer documentation and start scoping a skill the way Boomi did: real configuration patterns, real error handling, real terminology. This is not a rebrand of your existing help center. It's a new artifact built for a new reader.
Publish it under an open license and make it easy to find from your developer site. The goal isn't to lock anyone in. It's to be the most useful, most accurate source an agent can reach for when someone asks about your category.
AI search visibility used to mean getting cited. It increasingly means getting used. Agent Skills are how you make that happen on your terms instead of leaving it to chance.
Works Cited
"Boomi Companion Overview." Boomi Developer Documentation, Boomi, 2026, developer.boomi.com/docs/BoomiCompanion/Boomi_companion_overview.
"Boomi Companion: Equipping the Agents Your Teams Already Use." Boomi Developer Documentation, Boomi, 11 May 2026, developer.boomi.com/blog/boomi-companion-equipping-agents-your-teams-already-use.
Dotson, Kyt. "Data Activation and Governance Was the Heart of Boomi World 2026." SiliconANGLE, 28 May 2026, siliconangle.com/2026/05/28/ai-data-activation-boomiworld.