Anthropic's Velocity Is the Real Story: What Their March Sprint Signals
TL;DR
Anthropic shipped 74 releases between Feb 3 and Mar 24, 2026. That is more than one per day across four product lines
The breakdown: 28 Claude Code, 15 Cowork, 18 API/infrastructure, 13 model/platform. This is not a chatbot company. It is a platform company
If you are choosing where to build AI delivery systems, the velocity tells you everything. Anthropic is building the stack for autonomous agent work
74 releases in 52 days.
That's not a typo. Between February 3 and March 24, Anthropic shipped 74 product releases across their entire stack.
Pawel Huryn at Product Compass tracked every single one. The breakdown tells the story better than any blog post could.
- 28 Claude Code releases
- 15 Cowork (desktop automation) releases
- 18 API and infrastructure releases
- 13 model and platform releases
That's more than one release per day. For nearly two months straight.
This is not a chatbot company
The number is interesting. The pattern is what matters.
Look at the breakdown again. Claude Code got the most attention. But Cowork, API infrastructure and the model platform all got significant investment.
This is not a company iterating on a chat interface. This is a company building an operating environment.
Claude Code is evolving into an agent runtime. Cowork handles desktop automation. The API layer provides the infrastructure for external integrations. And the model platform keeps improving the intelligence underneath everything.
All four product lines advancing at once. In parallel. At speed.
What 74 releases tells you about strategy
Three things jumped out to me.
1. Velocity is the moat.
Most AI companies talk about model quality. Anthropic is competing on speed of iteration. 74 releases in 52 days means they're identifying gaps, building fixes and shipping them faster than anyone can keep up with.
For teams building on this stack, that means problems get fixed fast. Features arrive fast. The platform improves while you sleep.
2. They're building for agents, not humans.
TechCrunch reported on the auto mode release on March 24. But auto mode is just one piece. The March sprint also included: /loop (run for up to 7 days unattended), dispatch (always-on persistent agents), remote control via mobile, scheduled cloud tasks and channel integrations.
Every one of these features serves the same purpose. Let agents run longer. With less human intervention. More reliably.
Builder.io's roundup called it "maturing from coding assistant into environment for operating agents over longer stretches."
That's exactly right. The destination is clear. Autonomous agents that execute multi-day workflows without someone watching them.
3. The platform question is settling.
If you're deciding where to build AI delivery systems, velocity matters as much as model quality. A platform that ships 74 improvements in 52 days will outpace one that ships quarterly updates. Not because any single release is better. Because the cumulative improvement compounds.
The Claude Code sprint
Let me zoom in on Claude Code specifically. Because that's the product I use daily with clients building AI delivery stacks.
In March alone:
- Auto mode lets Claude Code work without asking permission at every step
- Computer use means it can interact with your screen and external applications
- /loop runs tasks for up to 7 days unattended
- Dispatch creates always-on persistent agents
- Remote control lets you manage agents from your phone
- Scheduled tasks run on a timetable without manual triggers
- Channels connect agents to Telegram, Discord and Slack
- 64K output tokens means longer, more detailed responses
- 1M context (now GA) means agents can process entire codebases
Each feature on its own is useful. Together, they add up to something different. A tool that can run your delivery processes while you're doing other work.
For non-technical founders, this changes the maths completely. You can now build custom AI delivery systems that operate autonomously. Not just respond to prompts. Actually run.
What this means if you're building AI into your business
If you're a consultant or educator thinking about which AI platform to build on, the velocity tells you everything.
Anthropic is investing in the infrastructure for autonomous agent work. Every release pushes in the same direction: agents that run longer, handle more complexity and need less hand-holding.
That's the stack you want underneath your IP.
But the stack alone doesn't create value. A powerful engine with no fuel is just expensive noise. The fuel is your structured methodology, your extracted IP, your delivery frameworks in a format that agents can operate from.
AI orchestration is what connects your IP to the platform. It's the layer between what you know and what the technology can do.
The platform is getting better every day. 74 times better in 52 days, apparently. The question is whether your IP is ready for it.
Want to find out? Take the assessment. Five minutes. You'll see how ready your IP is for the platforms being built right now.
Sources:
Frequently Asked Questions
James Killick
Founder
The AI Orchestrator. 10+ years building digital products and 200+ apps shipped, now helping $1M+ educators and consultants turn their IP into AI-powered delivery systems.
James Killick founded and runs The AI Orchestrators.
More from James Killick