Most people have heard of AI agents and haven't used one yet. We use them every day, and this website is built and updated by one working alongside Steve. Here's the plain version.
On November 5, 2026, Steve joined KXAN's Will DuPree and Avery Travis to talk about AI agents, ahead of Ragan's Future of Communications Conference in Austin.
WATCH ON KXANFrame: KXAN Austin. Plays on kxan.com.
A chatbot answers a question. An agent takes a goal and works on it.
You tell it what you want done. It figures out the steps, uses the tools it has access to, like a browser, your files or other apps, and keeps going until the job is finished or it needs you to make a call.
It's closer to handing work to a capable new assistant than to typing into a search box. You still decide what the work is and what good looks like.
In creative work, agents now show up in two places.
Inside the tools. Platforms like Higgsfield, Runway and Luma have agents built in. You describe what you want, and the tool handles the production steps.
Above the tools. An orchestration layer, like Claude's Cowork, plans the whole job and calls the right tool for each step. It's the difference between one person doing every job on a set and a director calling the shots.
The second level is the one that changes how much one person can make.
This website is the example we can show you. An AI agent writes and edits the pages, checks how they look on screen, pulls search data, and publishes the updates. Steve sets the direction, makes the calls and reviews the work.
We use the same approach to build AI video and content systems for brands and studios, where agents handle the repetitive production steps and people handle the story.
Agents are fast, but they don't have judgment. They'll confidently do the wrong thing if you let them.
Give an agent access to only what the job needs. Keep a person approving anything that sends, spends, publishes or deletes. Check its work the way you'd check a new hire's, and expect to catch mistakes.
Our own agent can't change our domain settings, for example. Steve does that himself.
A Studio of One, November 18. Steve's hands-on Austin AI Alliance class on directing a whole crew of AI agents to make video, podcasts and infographics in one evening. Details and tickets
For your team. Workshops and executive sessions on putting agents to work in communications and marketing. Speaking and workshops
An AI agent is software you give a goal to, rather than a single question. It works out the steps, uses tools like a browser, your files or other apps to carry them out, and checks back with you when it needs a decision.
A chatbot answers what you ask. An agent does the work: it can research, write, edit files, build things and move between apps until the job is done. You still decide what the job is and approve what matters.
For the work around the work: research, first drafts, turning one piece of content into many formats, building pages and reports, and keeping track of what's been done. The judgment about what to say and to whom stays with people.
It's one agent directing the others. Creative tools like Higgsfield, Runway and Luma now have agents built in. An orchestration layer, like Claude's Cowork, sits above them, plans the work and calls the right tool for each step, the way a director runs a set.
They can be, with limits. Give an agent access only to what the task needs, keep a person approving anything that sends, spends, publishes or deletes, and check its work the way you'd check a new hire's.
They change jobs more than they erase them right now. Tasks that were tedious get faster, and the value moves to the people who know what good looks like and can direct the work. That shift is real and it's worth preparing your team for.
Steve Mudd teaches A Studio of One, a hands-on Austin AI Alliance class on directing a crew of AI agents to make video, podcasts and infographics, on November 18, 2026. He also runs workshops for teams.
We'll show you where they help and where they don't. We're in Austin.
Let's talk