I wrote about Claude Dispatch yesterday, which is by all measures an awesome product. My personal conclusion was that it’s early for its time and there are limitations surrounding it. Specifically, it doesn’t gel well with my corporate laptop. And that’s where my actual heavy jobs live.

But after reflecting on my observations yesterday, I think it makes total sense for Anthropic to pivot their product strategy and start targeting the business segment. Cowork and Dispatch are potentially signals of where Anthropic is heading: enterprise AI.

Virality and revenue are different things

Great consumer products don’t automatically translate to great businesses. There’s a big difference here. Firstly, consumers consume content. Secondly, companies pay for productivity. They’re totally different customer jobs, and resultantly different monetisation opportunities.

The flywheel (I’m currently reading Jim Collins’s book Good to Great) goes like this:

• Frictionless consumer adoption
• Organic enterprise discovery (people use it at home, then want it at work)
• Enterprise is where willingness-to-pay actually lives

Targeting businesses is a better bet for revenue in any case. A consumer pays $20/month for a Claude subscription; a business, on the other hand, pays per seat, often at a higher tier, across tens or hundreds of employees. Add bundled products charged at higher tiers for businesses and you can up-sell and cross-sell.

On top of that, businesses have budgets for productivity tools. Consumer subscriptions are typically a discretionary spend and the first thing to get cancelled when money is tight. Businesses also have a stronger reason to pay. If Claude saves a knowledge worker two hours a day, that’s real money back to the company.

Equally important, consumer AI is mostly Google and Meta’s territory. They have a large enough user base to monetise through ads. For Anthropic, which is new to the game, consumer subscriptions alone won’t carry the business long-term. I mean, just look at what happened to OpenAI’s experiment with Sora last week. So the pivot to enterprise makes sense for Anthropic.

The Figma playbook shows how the flywheel works in practice

Figma is a textbook case (and one I love) of a bottoms-up motion for product adoption. It starts off with one designer adopting it in an organisation. They then share a file internally. That file pulls in collaborators (for free). And every shared file becomes a distribution event. In other words, the product is able to sell itself before procurement ever gets involved.

Then lock-in compounds quietly:

• Design systems
• Component libraries
• Institutional memory

Deep workflow integration creates switching costs that compound over time. Once a product is load-bearing inside how a team works, removing it is a people problem, not a software problem.

Now look at what Anthropic is building:

• Dispatch (async task execution from your phone)
• Cowork (persistent desktop workspace)
• Claude Code (agentic coding)
• MCP (protocol for connecting AI to enterprise tools and data)

Those aren’t standalone features. They’re the underpinnings of enterprise integration. Each feature pushes Claude deeper into how people work. And agentic workloads embedded in enterprise processes are the strongest lock-in there is.

Anthropic has a real opportunity to tap into revenue here by pursuing the business segment. It’s the same exact strategy that Microsoft has focused on in the past decades.