Bots and Skills and Oh My!

AI agents aren’t just getting smarter — they’re getting skilled and can actually do a ton. It seems like just yesterday we were coding data-driven websites on notepad or written some pretty cool formulas in Excel but now we have entered a different era. The AI bots are now "hands on keyboard" and we've moved beyond the era of "asking questions" to actually doing stuff. AI is doing real work, and today, getting AI to work for me has come in handy and has been completely easy to do. However, I can see how it is easy for a novice in tech, AI, cloud, systems, and security can make a ton of mistakes. In my other blog, we talked about Clawdbot, Kimi, and Moltbook doing some crazy things. Well, the "skills" are the things - like a buffet where you can eat whatever you want.

In the last couple of days, I have hired a workforce that is managing my inbox and unsubscribing me from junk crap and getting me organized and to net zero emails, while updating my GitHub and Hugging Face Repos, and checking on my mom in her care facility 1500 miles away. Checkout my GitHub and my Hugging Face Reachy "Check On Mom" app I created to literally , literally wake-her up with my voice, play her favorite tune list I curated, and monitor her vitals and her prescriptions. Soon we will be playing Canasta with my sister and her virtual best friend. AI agents like Clawdbot are entering a new era where they can be extended just like smartphones were with apps. Instead of hard‑coding integrations, users can now install Skills—standardized instruction files that teach an AI how to use specific APIs, tools, or environments. A Skill acts like a capability pack: it tells the agent what a tool does, how to call it safely, and when to use it to complete a task. If you look at just a fraction of what Skills can do, and then put them all together, this can be a good thing or a bad thing. Do I want that "Smart" anything or is this where we unplug as much as possible and hope we implement the right controls and continuous monitoring.

Productivity & Workspace Skills

These Skills turn the agent into a full digital assistant that can actually do things, not just talk about them. Already we are going back to bots now executing scripts we used to write, but now the bots are doing that.

Developer & DevOps Skills

This is where AI becomes a junior engineer that can execute real actions. I used to code for a while until .NET came around and then I switched to cybersecurity. Now, I am a developer, architect and tester all at once. Of course, AI isn't perfect and I am glad I can code Python and JavaScript to fix the error of AIs ways, but most of it.... I've amazed myself!

Data and Analysis Skills

I used to love data warehousing back in the day. Taking large datasets and writing my own ETL and parsing and cleansing routines. When I was a trainer I used to tell my students to by a book called "T-SQL Puzzles and Games" and do challenges on complex statements. Now, again, I can do all sorts of data activities. Query any kind of database, place it into any kind of format and create all associated meta-data around it, to do more things with it.

4. Automation & Integration Skills

This is where AI becomes the glue across an entire tech stack. It brings so manty different things together that really let us know what sort of workflows are possible and how we can bridge together so much.

5. Smart Home & IoT Skills

A Skill can turn the agent into a home automation controller. Look at what Reachy is doing to homes with Smart TVs and fridges, and check out our learning tools in our education section.

6. Custom Enterprise Skills

Companies can build skills that do a lot for the business, but this is where safe skills become a reality.

Skills Are a Big Deal

Skills shift AI from “chatting” to doing. They give users the power to extend an AI agent’s abilities the same way apps extended the iPhone. And because Skills are standardized, portable, and easy to install, they create an ecosystem where anyone can teach an AI to use any tool. This is the moment where AI stops being a passive assistant and becomes an active operator.