Anthropic is teaching developers for free while tightening how customers pay. The company published a free Claude Courses library on Skilljar, from 15-minute primers to an eight-hour, 85-lecture "Claude with Amazon Bedrock" series; a reviewer said they completed a 20-minute "Introduction to subagents" lesson in roughly 20 minutes. At the same time Anthropic changed Claude Code billing for third-party harnesses, launched a routines research preview, redesigned its Mac app, and announced a SpaceX compute deal that raises internal usage caps, moves that affect how developers learn, build, and scale with Claude.
Anthropic is pushing two tracks at once. On one track it's teaching users how to use Claude and Claude Code for free. On the other it's reshaping how the product is packaged and billed.
Free courses, short commitments
The company published a library of Claude Courses on Skilljar that run the gamut from bite-sized primers to deep technical series. The shortest module named in Anthropic's catalog is a 15-minute "AI Capabilities and Limitations" primer. At the other extreme a multi-hour course, "Claude with Amazon Bedrock," runs eight hours and spans 85 lectures. Registration, progress tracking, and course access on Skilljar are free.
A ZDNET reviewer tested the middle of that range. The reviewer said they completed the 20-minute "Introduction to subagents" module in roughly 20 minutes. The lesson contains four short YouTube videos and supplemental text. This reviewer reported practical takeaways, including guidance on when to use subagents, how subagents can leverage Claude's context window, and how to specify discrete tasks for subagents. ZDNET also noted the course issues a completion certificate.
That mix is designed for different user needs. Someone who wants a quick primer can finish a lesson in under half an hour. Engineers or teams that need platform integrations can sign up for the longer, multi-lecture offerings. All of it sits behind a free sign-up on Skilljar, so Anthropic is lowering the barrier to entry for developers and product teams who want hands-on familiarity with Claude's tooling.
Product shifts: routines, Mac app, and third-party billing
At the same time Anthropic is changing the operational side of Claude Code. The company released a research-preview feature called routines that runs scheduled, repeatable automations on Anthropic's web infrastructure. According to 9to5mac, routines package access to repositories and connectors so they can run scheduled tasks, API workflows, and GitHub routines even when a developer's Mac is offline.
9to5mac also reported the company published daily execution limits for routines tied to plan tiers. Pro accounts can run five routines per day. Max accounts can run 15 per day. Team and Enterprise accounts can run 25 per day. Those caps are part of the preview rollout and shape how teams might rely on Anthropic's hosted automation rather than local devices.
The Claude Code experience in Anthropic's Mac app also got a refresh. 9to5mac describes a redesigned interface that supports multiple parallel sessions, a sidebar for session management, an integrated terminal, built-in file editing, HTML and PDF preview, and an improved diff viewer. The changes aim to make Claude Code feel more like a native developer environment on macOS.
But not all third-party integrations will remain covered by subscription limits. TechCrunch reported that starting at noon Pacific on April 4 Anthropic told subscribers they would no longer be able to use their Claude subscription limits for third-party harnesses such as OpenClaw. Those harnesses must now be paid for through a separate pay-as-you-go option billed outside the subscription.
TechCrunch reported the company framed the move as a response to engineering constraints and atypical usage patterns for these third-party tools. Boris Cherny, head of Claude Code, wrote on X that "subscriptions weren’t built for the usage patterns of these third-party tools" and that Anthropic was trying "to be intentional in managing our growth to continue to serve our customers sustainably long-term." TechCrunch also said Anthropic is offering full refunds to subscribers who were unaware that third-party harness usage was unsupported under their subscription.
The change drew pushback. TechCrunch cited a comment from Peter Steinberger, the creator of OpenClaw, who joined OpenAI and accused Anthropic of "locking out open source." Anthropic disputed the characterization and emphasized engineering limits as the reason for the new billing approach.
Those billing changes coexist with a separate capacity expansion. At its Code with Claude developer conference on May 6 Anthropic announced a capacity deal with SpaceX intended to expand onshore compute and lift internal usage caps. Ars Technica reported the arrangement gives access to more than 300 megawatts of compute capacity and highlighted SpaceX's Colossus 1 supercomputer as part of the deal. Ars Technica also reported that SpaceX said Colossus 1 includes "over 220,000 NVIDIA GPUs, including dense deployments of H100, H200, and next-generation GB200 accelerators."
Anthropic's CEO Dario Amodei framed the SpaceX relationship as a way to raise Pro and Max plan usage limits for Claude Code. Ars Technica noted Anthropic posted changes that doubled prior five-hour window limits for Pro and Max subscribers, removed peak-hours limit reductions for those accounts, and raised API limits for its Opus model. Those moves expand internal capacity while Anthropic reorganizes how third-party tooling consumes subscription resources.
The company told customers the third-party harness billing policy "applies to all third-party harnesses and will be rolled out to more shortly." Anthropic also said it has expressed interest in working with SpaceX to expand into orbital compute capacity, though the company didn't give a timetable.
Related Articles
- AI logo generators: 5 prompt-based tools and what they deliver
- Amazon Now: 30-Minute Delivery Hits Dozens of U.S. Cities
- TV Writers Take AI Labeling Jobs, Report Pay Up to $150
Anthropic says the SpaceX capacity deal provides additional onshore compute capacity and that it has adjusted Pro and Max usage windows. The company also published routines preview limits: five routine executions per day for Pro accounts, 15 for Max, and 25 for Team and Enterprise accounts.
This article was created with AI assistance.