Everyone's an agent now

We don't compete in marketing automation. ORCA HQ sells governed knowledge infrastructure for professional-services firms; the marketing automation category is a different game with different incumbents. That makes us a useful kind of observer — close enough to the AI-application stack to read the signals, far enough from the fight to say what we see without selling against any of the players named below.

This piece is what May 2026 looks like when you read the hero lines of the major marketing automation vendors back-to-back. We pulled them off the live homepages on 3 May 2026; quotes are verbatim from those pages and linked at the bottom. Three patterns jumped out immediately, a fourth is the most interesting one, and a fifth is the one buyers should pay closest attention to.

1. The agent metaphor has won the messaging war

Every major marketing automation vendor has at least one named "AI agent" product on the homepage today. Different brand vocabulary, identical conceptual product:

HubSpot ships Breeze — including a "Breeze Customer Agent" and a "Blog Research Agent" — on a hero line that reads "AI-Powered Marketing Software that Multiplies Results." Klaviyo calls itself "the autonomous B2C CRM" with a Marketing Agent and a Customer Agent under the K:AI banner ("Customer Agent resolves 65% of questions autonomously"). Braze opens with "Rewrite the rules of customer engagement with AI" and ships BrazeAI™ Agents alongside a Decisioning Studio. Iterable's hero is "Most platforms react. Iterable adapts intelligently and in real time." — backed by Nova Insights, Nova Decisioning, and Nova Agents. ActiveCampaign calls it "Active Intelligence." Customer.io simply says "Meet your AI Agent."

Twelve months ago the category-default was "marketing automation platform with AI features." In May 2026 the category-default is "AI agents for marketing." The agent has become the noun marketers buy. Anyone whose homepage still leads with "automation" in May 2026 reads as a generation behind the conversation, and the market knows it.

The agent isn't a feature any more. The agent is the product. The platform is what the agent runs on. That inversion — feature becoming product, product becoming substrate — is the largest single shift in software-category positioning since the move from on-premise to SaaS.

2. "Autonomous" has quietly replaced "automated"

Klaviyo: "the autonomous B2C CRM." ActiveCampaign: "autonomous marketing." Braze: "always-on AI agents" doing "AI-powered decisioning." HubSpot: "Put tedious marketing tasks on autopilot."

The word matters because the contract changes. Automation means: the operator builds rules, the system executes them, deviation requires the operator to update the rules. Autonomy means: the operator sets goals and guardrails, the system chooses tactics within them, deviation is by design.

Buyers who sign for "autonomous marketing" in 2026 are signing up for a different operating model than buyers who signed for "marketing automation" in 2024. The platform is now allowed to make decisions the buyer didn't pre-approve. That's a feature for teams that have the discipline to set good guardrails and the observability to notice when the system drifts. It's a liability for teams that don't. The category messaging hasn't yet caught up to that distinction; the buyer questionnaire needs to.

3. MCP just reached the marketing automation category

The most quietly significant signal in the survey: Customer.io's May 2026 homepage now positions itself as "Built for agents, not just humans. Works with any AI tool using MCP, CLI, API or webhooks." That's the Model Context Protocol — the open standard for AI tool connectivity — being named on the marketing-automation homepage of a mainstream platform that talks to non-engineering buyers.

MCP first showed up as a developer-tool concern in late 2024. Eighteen months later it's category vocabulary on a marketing platform's hero section. That's the typical path for protocols that win — they start in developer surfaces, get adopted by infrastructure layers, and eventually get named in product positioning when buyers start asking for them. We're at the third stage now in a category that wasn't expected to care.

The implication for buyers across categories — not just marketing automation — is that MCP-native is becoming a procurement question buyers can ask out loud. Vendors without an answer will spend 2026 figuring out one. Vendors with one will get to use it as a wedge. The question on the buyer side is no longer "do you have an AI feature?"; it is "what protocol does your AI feature speak?"

4. The language is converging — and one phrase is appearing verbatim

This is the observation we found most interesting and have to be most careful about saying. Customer.io's homepage carries this line: "Intelligence that compounds. The more you use the platform, the sharper it gets."

The first half — "Intelligence that compounds" — is the line ORCA HQ has used in our hero, in our footer, on our about page, in our pricing copy, and in our blog tagline since we launched in March 2026. We didn't trademark it. We don't claim Customer.io copied us; we have no evidence of that and they're a much larger company with an independent marketing team. What we observe is that the same phrase has emerged on two unrelated companies' homepages in the same quarter, expressing the same idea.

This happens when a category is converging on its eventual vocabulary. The first generation of any new technology category produces wildly diverse positioning language; the survivors converge on the words that actually capture the value, and those words become the category's grammar. "Software as a service" was a synonym wrestle for years before it became SaaS. "Customer relationship management" had a dozen alternative framings before it became CRM. "Intelligence that compounds" may or may not be one of those phrases — it's too early to tell — but the convergence is the signal worth noticing.

For marketing automation specifically: the value-proposition vocabulary is converging on three families. Autonomous for what the system does. Agent for the user-facing product unit. Compounding / improving / sharpening for the trajectory promise. A vendor whose hero in 2027 doesn't sit somewhere in that triangle will read as out-of-category.

"When the same phrase shows up on two unrelated vendors' homepages in the same quarter, it's not theft. It's the category settling on its grammar. The next phase is when that grammar becomes the procurement questionnaire."

5. The privacy gap is enormous — and it's the buyer's leverage

Klaviyo's December 2025 trends piece named "Privacy and consent will drive personalization" as the third of eight 2026 trends, and analyst writeups from the same window all flag privacy-first personalization as a defining theme. None of the homepages we surveyed put privacy in the hero. Compliance badges (SOC, ISO 27001, GDPR) appear on the trust pages and in the footer; privacy framing appears under "trust centre" links. None of it is the front door.

This is the opposite of what the trend coverage says is happening. The category is talking about privacy in long-form content (blog posts, trends pieces, whitepapers) but selling on capability and outcomes in the hero. The gap is structural — and for buyers it's leverage. A vendor who can credibly lead with governance-first positioning in 2026 has a wedge no incumbent has yet stepped on. The procurement questionnaire that asks "show me Article 13 transparency evidence" before "show me ROI" is the questionnaire incumbents are not yet built to answer.

If you're on the buyer side of a marketing automation procurement in May 2026, the most differentiating question you can ask isn't "what AI agents do you ship?" — every vendor will answer that one fluently. It's "what does your platform refuse to do, in code, that a less governed system would do by default?" The answers separate the architectures from the marketing.

6. Outcome-claim inflation is universal — and that's a buyer signal too

Across the surveyed homepages: HubSpot reports "129% more leads, 36% more deals after 1 year." Braze cites a Forrester study of "457% ROI and payback in less than six months." Klaviyo highlights "65% YoY growth" for one customer. Iterable cites "300% increase on revenue." Customer.io shows "28% Increase in click-through rates."

Every vendor is winning. Every vendor's customers are winning. None of the numbers are independently sourced beyond the vendor's own case studies; the Forrester study cited by Braze is the sole exception, and Forrester studies commissioned by vendors carry their own well-known interpretation caveats. None of this means any specific number is wrong — it means the headline numbers are not where buyer due diligence should land.

The numbers worth asking about in 2026 are the ones the homepages don't carry: time-to-first-value, average procurement cycle length, churn after year one, the percentage of customers who actively use the AI agent features versus the percentage who configured them once and forgot, the support-ticket-to-customer-success-conversation ratio. These are the numbers that show whether the autonomous-marketing pitch is working in production or whether it's a vibes-driven pilot that hasn't survived a real renewal cycle.

What this is and isn't

This isn't a competitive hit piece. The marketing automation category is doing serious work and the named vendors are credible operators. The convergence we observe is healthy — categories that don't converge on shared vocabulary fragment into incompatible niches and stop growing. The autonomous-agent pitch isn't wrong; it's the right next chapter for what marketing automation has been trying to be since the late 2010s.

It also isn't an analyst report. We don't have access to the Q1 2026 earnings figures, the Gartner Magic Quadrant data, the Forrester Wave scores, or any specific seller's ARR breakdown. The analysis above is what's observable from the public messaging surfaces of the vendors named, on a single specific date. If you want the financial-rigour version, talk to the analysts who do that work.

What this is: a snapshot of how the category is positioning itself right now, by a writer who isn't trying to win against any of these vendors. Categories tell you most about what they're going to be by what they all start saying at the same time. May 2026 the marketing automation category is saying "agents," "autonomous," "MCP," and "compounding." The buyers, the regulators, and the next-wave entrants should plan accordingly.

Where ORCA HQ sits — honestly

For completeness: ORCA HQ doesn't ship marketing automation. We sell governed knowledge infrastructure for professional services firms — different category, different buyer, different value proposition. The platform is MCP-native (so Customer.io's MCP positioning is a category we're already inside). The discipline is governed orientation (which is why we've been using "intelligence that compounds" as our line since launch). The agents we run are infrastructure agents — heartbeat, auditor, gardener, scout, prospector — not the customer-facing autonomous agents the marketing automation vendors are now branding.

If you're on the buyer side of any AI-application procurement in 2026 — marketing automation, knowledge management, support, or any adjacent category where the same patterns are visible — the questions worth asking are the ones we listed above. Talk to the vendors you're evaluating about them. Talk to us if you want to think them through with someone who isn't selling you the answer.

Sources. Quotes above pulled from these vendor pages on 3 May 2026:

Quoted figures (lead growth, ROI, retention) are vendor self-reported on the linked pages and have not been independently verified. Forrester study cited by Braze is a vendor-commissioned analyst report.

Everyone's an agent now.