Anthropic just made a major play in the enterprise AI race with Skills for Claude - customizable folders packed with instructions that teach the AI to master specific workplace tasks. The move puts pressure on OpenAI after its recent AgentKit announcement and signals the battle for practical AI agents is heating up across corporate America.
The AI agent wars just got more practical. Anthropic dropped Skills for Claude on Thursday, marking its most aggressive push yet into workplace automation territory that OpenAI has been eyeing with increasing intensity.
Skills transforms Claude from a general chatbot into something closer to a personalized workplace assistant. Think of it as instruction folders that load automatically when you need them - whether you're wrestling with Excel spreadsheets or following your company's specific brand guidelines. The feature works across Claude.ai, Claude Code, Anthropic's API, and the Claude Agent SDK.
"The thing that's interesting to me about Skills is basically about agents," Brad Abrams, a product lead at Anthropic, told The Verge. He emphasized this isn't about hitting benchmark numbers - it's about getting real work done at your actual company.
The timing isn't coincidental. This comes weeks after OpenAI unveiled AgentKit at its annual DevDay event, complete with a flashy demo showing how Albertsons could use custom agents to boost ice cream sales. Both companies are racing to crack the same puzzle: making AI agents actually useful instead of just impressive in demos.
Early enterprise customers are already testing the waters. Box, Canva, and Rakuten have been putting Skills through its paces, according to Anthropic's release. Abrams demonstrated the potential by using Claude's PowerPoint Skill to generate a presentation about Haiku 4.5's market performance, creating "well-formatted slides that are easy to digest."
This represents a shift from the AI industry's obsession with benchmarks toward solving actual workplace friction. Instead of spending time crafting the perfect prompt every time you need something done, Skills lets you front-load that context and training.
The broader competitive landscape reveals just how high the stakes are. Google, Microsoft, OpenAI, and Anthropic have been pouring resources into agentic AI for years, with executives regularly hyping the technology on earnings calls. But progress has been frustratingly incremental - think Anthropic's Computer Use feature or OpenAI's parade of agent announcements like Operator, Deep Research, and ChatGPT Agent.
What's different now is the focus on enterprise deployment rather than consumer wow factor. Skills targets Pro, Max, Team, and Enterprise users specifically, while OpenAI's AgentKit similarly aims to help companies "take agents from prototype to production." Both companies seem to have realized that the path to AI agent adoption runs through corporate IT departments, not viral TikTok demos.
The enterprise angle makes sense financially too. Companies are willing to pay premium prices for AI tools that measurably improve productivity, while consumer AI remains challenging to monetize beyond subscription fees.
But challenges remain. AI agents still struggle with complex, multi-step workflows that require real judgment calls. They work best with clearly defined, repeatable tasks - exactly what Skills appears designed to handle.
Skills represents Anthropic's bet that the future of AI agents lies in practical workplace integration rather than flashy demos. By letting companies customize Claude for their specific needs, Anthropic is positioning itself as the enterprise-friendly alternative to OpenAI's more consumer-focused approach. The real test will come as more companies deploy these tools and discover whether AI agents can finally bridge the gap between promise and productivity.
