61% of marketers say this is the biggest marketing shift in decades.

Get the data and trends shaping growth in 2026 with this groundbreaking state of marketing report.

Inside you’ll discover:

Results from over 1,500 marketers centered around results, goals and priorities in the age of AI

Stand out content and growth trends in a world full of noise

How to scale with AI without losing humanity

Where to invest for the best return in 2026

Download your 2026 state of marketing report today.

Get Your Report

Anthropic just commoditised the hardest part of building AI agents. On 8 April 2026, Anthropic launched Claude Managed Agents, a hosted service giving developers out-of-the-box infrastructure to build and deploy autonomous AI systems at scale. The product bundles an agent harness, a sandboxed execution environment, and cloud-based monitoring with permission controls, removing the distributed systems engineering problem that previously required dedicated teams to solve. Anthropic's annualised recurring revenue has now surpassed $30 billion, roughly triple its December 2025 figure, with the majority of that growth coming from its enterprise API platform. For teams building on AI, this matters because the barrier that separated companies with serious engineering resource from those without has effectively been lowered to a subscription. Notion is already using it to automate client onboarding. The engineering problem is solved. The product problem is now yours.

Canva is no longer a design tool. It is building the infrastructure for the entire marketing cycle. On 8 April 2026, Canva announced the simultaneous acquisition of Simtheory, an AI agent management platform, and Ortto, a customer data and marketing automation company serving over 11,000 customers across 190 countries. Both were founded by brothers Chris and Mike Sharkey, who will take leadership roles across Canva's AI and marketing technology teams. Simtheory adds agentic workflow capabilities; Ortto adds a full customer data platform with automation across email, SMS, push, in-app messaging, and surveys. For solopreneurs and small marketing teams already inside Canva, the implication is straightforward: the case for maintaining a separate marketing automation platform just got considerably weaker. Canva's COO described this as the shift from "a design platform with AI tools to an AI platform with design and productivity tools at its core." Stack consolidation is no longer a cost-cutting exercise. It is the product strategy.

Google's first-page rankings are nearly invisible inside the AI discovery layer, and most marketers have not noticed yet. Citation analysis published the week of 3 April 2026, drawing on Ahrefs and eMarketer data, found that only 8% of ChatGPT citations originate from Google's top 10 search results. Gemini is barely better at 8.6%. Perplexity is the outlier at 28.6%, but the more striking figures are on the platform side: Reddit accounts for 40.1% of all generative AI citations globally, followed by Wikipedia at 26.3% and YouTube at 23.5%. For marketers whose SEO strategy has been built around ranking on page one of Google, this is not a refinement problem. It is a category problem. The channel where you have been optimising is not the channel where AI systems are looking. If your brand has no Reddit presence, no Wikipedia entry, and no YouTube library, you are already absent from a significant share of AI-driven discovery.

AI martech vendors are shifting from capability claims to outcome accountability, and that changes every procurement conversation. In the week of 9 April 2026, HubSpot restructured its Breeze AI pricing to a pay-per-result model: its Customer Agent resolves 65% of support interactions autonomously with a 39% reduction in resolution time across 8,000 customers, and its Prospecting Agent is now priced at $1 per qualified lead. Separately, Adobe's 2026 AI and Digital Trends report, surveying 3,000 executives and 4,000 customers, found that 78% of organisations expect agentic AI to handle at least half of all customer support within 18 months, with the primary bottleneck identified as data fragmentation rather than model capability. For marketing budget owners, pay-per-performance pricing reframes the ROI conversation entirely. You are no longer being asked to believe in potential. You are being asked to verify results before the invoice grows. The vendors who cannot price that way are telling you something.

The pattern: What connected this week's stories was not any single product launch. It was the same structural shift appearing simultaneously across the infrastructure, platform, and measurement layers of marketing technology. Each story is about a different part of the stack, but they all point in the same direction: the experimental phase is closing, and the infrastructure phase has begun. The marketers who treat this as incremental will optimise for a world that is already being dismantled. The ones who treat it as structural will be building on the new foundations, while others are still arguing about whether they exist.

Sources: Claude Managed Agents launch · Canva acquires Simtheory and Ortto · AI citation sources analysis · HubSpot pay-per-performance pricing

Anthropic's Managed Agents didn't just launch a hosting service. It collapsed the gap between teams who can build AI agents and teams who can't.

A couple of weeks ago, a project called Paperclip was trending in the AI builder community. If you missed it, the idea was straightforward: an orchestration layer that helps you coordinate multiple AI agents working together as a team. Not one agent doing one task, but several agents collaborating, handing off work, checking each other's output.

I found it genuinely interesting. Not because the tool itself was polished, but because it pointed at a problem that was clearly bothering a lot of people. The hard part of AI agents in 2026 isn't the model. It's everything around the model. The harness. The execution environment. The monitoring. The permissions. The ability to let something run for hours without it falling over or doing something you didn't expect.

Then, on 8 April, Anthropic launched Claude Managed Agents, and I think that's their direct answer to the same problem Paperclip was solving, except productised, hosted, and backed by enterprise-grade infrastructure.

That launch matters more than it looks.

There's a belief I keep running into when I talk to marketing teams and founders about AI agents: that the thing holding them back is model capability. They're waiting for the model to get smarter. Waiting for it to handle more complex instructions. Waiting for the next version.

That's right, but it's not the whole picture.

The actual bottleneck for most teams, especially those without a dedicated engineering function, has been the infrastructure layer. Building an agent that works reliably in production requires distributed systems engineering. You need a harness that manages the loop between the model and its tools. You need a sandboxed environment where the agent can execute code safely. You need monitoring so you can see what it's doing. You need permission controls so it doesn't access things it shouldn't.

That's not a prompt engineering problem. That's a software engineering problem. And for the vast majority of marketing teams, solopreneurs, and small product companies, it was a wall.

What Anthropic did with Managed Agents is take that wall down. They decoupled the session, the harness, and the sandbox into swappable components, bundled them into a hosted service, and made it available through their API platform. The Notion demo showed an agent autonomously working through a client onboarding checklist, with a dashboard showing exactly what the agent was doing and which tools it was using.

The engineering problem is now a product you subscribe to. The question that remains is what you build on top of it.

Read the full article here…