Anthropic’s ‘Agent Skills’ Revolutionizes Workplace AI, Challenging OpenAI’s Standard

Anthropic recently announced that it will be releasing its Agent Skills technology as an open standard, a strategic move aimed at solidifying its position in the rapidly evolving enterprise software market. The San Francisco-based artificial intelligence company also introduced organization-wide management tools for enterprise customers and a directory of partner-built skills from companies such as Atlassian, Figma, Canva, Stripe, Notion, and Zapier.

The introduction of these new features represents a significant expansion of the technology that Anthropic first introduced in October. What started as a developer feature has now evolved into infrastructure that is poised to become an industry standard. According to Mahesh Murag, a product manager at Anthropic, they are launching Agent Skills as an independent open standard with a specification and reference SDK available on their website.

Skills are essentially folders containing instructions, scripts, and resources that enable AI systems to perform specific tasks consistently. This approach addresses the challenge of large language models lacking specific procedural expertise needed for specialized professional work. By packaging procedural knowledge into reusable modules, skills allow AI assistants to complete specialized tasks more efficiently.

Anthropic’s system is designed around the concept of progressive disclosure, where each skill only takes up a few dozen tokens in the AI’s context window. This allows organizations to deploy extensive skill libraries without overwhelming the AI’s working memory.

Enterprise customers are already utilizing skills in various functions such as legal, finance, accounting, and data science. The new enterprise management features enable administrators to provision skills centrally, controlling which workflows are available across their organizations while allowing individual employees to customize their experience.

Anthropic is launching with skills from ten partners, including industry leaders like Atlassian, Figma, Stripe, and Zapier. The company is focusing on ecosystem development rather than immediate revenue generation, with no revenue-sharing arrangements in place at this time.

The decision to release Skills as an open standard is a strategic one, aimed at fostering ecosystem growth. This move has already been adopted by other industry players like OpenAI, indicating a common approach to making AI assistants more capable at specialized tasks.

The Skills framework represents a shift in how the AI industry approaches making AI assistants more capable. Rather than building specialized agents for different use cases, organizations can invest in creating and curating skills that enhance their AI assistants’ capabilities.

However, the Skills framework also raises concerns about maintaining human expertise and potential security risks. Anthropic recommends installing skills only from trusted sources and conducting thorough audits of less trusted origins.

In conclusion, Anthropic’s release of Agent Skills as an open standard signifies a shift towards skills becoming infrastructure in the AI industry. The expertise organizations encode into skills today will determine the effectiveness of their AI assistants in the future, regardless of the model powering them. The industry has converged on a common approach to making AI assistants more capable, and it came from the company that chose to give it away.