AI Agents Are Evolving From Experiments to Real Workplace Helpers — But Here's Where It Gets Controversial...
For years, AI agents existed mostly as a concept or experimental tool. Companies have poured resources into creating them, but the real challenge has always been making them genuinely useful for day-to-day tasks, whether for individual consumers or professionals. Now, the focus is shifting from theory to practical application.
Anthropic took a major step in that direction on Thursday by introducing Skills for Claude. This feature allows Claude to access specialized "folders" containing instructions, scripts, and resources, which enhance its ability to handle specific work-related tasks. From managing Excel spreadsheets to following an organization’s unique brand guidelines, Skills aim to make Claude far more competent in practical scenarios. Users can even design their own Skills tailored to their roles and deploy them across Claude.ai, Claude Code, Anthropic’s API, and the Claude Agent SDK. Companies like Box, Rakuten, and Canva have already experimented with the tool.
At its core, this feature is designed to make Claude a more capable AI assistant for your specific work needs. Instead of constantly refining prompts or digging through previous context, you can rely on Skills to streamline the process. Currently, the feature is available for Pro, Max, Team, and Enterprise users.
Brad Abrams, a product lead at Anthropic, told The Verge that the appeal of Skills lies in empowering agents. He explained that Skills give organizations the ability to train Claude to perform tasks effectively within their unique work environment. This is not about hitting arbitrary AI benchmarks — it’s about enabling Claude to complete real tasks for real companies.
Abrams gave a concrete example: by applying an Anthropic layer on top of Claude’s PowerPoint Skill, he had Claude generate a presentation about the market performance of Haiku 4.5. According to him, Claude produced slides that were well-structured and easy to understand — a practical outcome that saves users time and effort.
For years, major tech players including OpenAI, Anthropic, Google, and Microsoft have been chasing the dream of truly useful AI agents. Executives regularly highlight the promise of agentic AI during earnings calls, shifting internal focus and resources toward building these tools. However, the progress has been mostly incremental, with companies rolling out feature updates or new iterations of their agents — like Anthropic’s Computer Use, or OpenAI’s evolution from Operator to Deep Research and finally ChatGPT Agent, which merges multiple capabilities.
Anthropic’s announcement follows OpenAI’s recent reveal at their annual DevDay. OpenAI introduced AgentKit, a suite of tools aimed at helping developers and large companies move AI agents from prototype to production. One striking example was Albertsons, a US grocery chain with over 2,000 stores, using a custom agent to create strategies for boosting ice cream sales when they dipped by over 30 percent. Other companies like Box, Canva, Evernote, and Ramp have also tested AgentKit. OpenAI additionally rolled out a consumer-focused tool that allows people to interact with apps such as Zillow and Uber Eats directly inside ChatGPT.
So here’s the thought-provoking question: as AI agents like Claude and OpenAI’s tools become increasingly capable in professional settings, will we start relying on them for critical decisions, or should humans remain firmly in control? Share your thoughts in the comments — this is where opinions are likely to diverge sharply.
