What's up, everyone!
I've spent a lot of time building software with Claude and other AI tools over the last couple of years, and I'll admit something. Sometimes I find myself arguing with a robot like it's an employee who just completely ignored everything I asked it to do.
These systems are incredibly smart, but they can also make some unbelievably dumb decisions. I'll give an AI agent what I think are clear instructions, it goes off and does something completely different, and suddenly I'm spending more time fixing the mess than I would have spent doing the work myself.
That's why something Joe Vaccaro told me really stuck.
Joe is an SVP at Cisco, where they're deploying AI agents across a company of roughly 90,000 employees. They're dealing with this technology at a scale most of us can't even imagine.
And when I asked him what advice he'd give business leaders trying to figure all this out, his answer was pretty simple.
Stop outsourcing your understanding of AI. Start using it yourself.
I couldn't agree more. We're moving into a world where knowing how to use these tools isn't just a technical skill. It's becoming a leadership skill. And the people who understand what AI can actually do, where it fails, and how to apply it to their businesses are going to have a serious advantage.

Here's What I Learned This Week
01If You're Leading a Company, You Need to Be Using AI Yourself
The Idea: Joe made a point I wish more executives understood. You can't fully appreciate what AI can do for your business by sitting through presentations about it or asking someone else to figure it out.
At Cisco, they've made AI agents available across approximately 90,000 employees. But Joe's recommendation for leaders is much more personal: build something yourself, even if it's just a weekend project.
Once you've experienced what these tools can accomplish, you start recognizing opportunities to change how your business operates.
Your Move: Block out two hours this weekend to use AI for a real business problem. Build a simple tool, automate a repetitive task, or create a process that saves your team time. The goal isn't perfection. It's firsthand experience.
02Your AI Agents Need Management, Not Blind Trust
The Idea: Joe shared an analogy that made me laugh because it's painfully accurate.
Think of AI agents like teenagers. They're confident, eager to get things done, and sometimes lack good judgment.
When you give an agent a goal, it may pursue that goal in ways you didn't anticipate. That's why guardrails, testing, permissions, and human oversight are essential, especially when agents can take actions inside your business.
Your Move: Before giving an AI agent control over an important workflow, define exactly what it can do, what it cannot do, and when it needs approval. Start with low-risk tasks and expand its responsibilities only after you've tested the results.
03The Better Your Instructions, the Better Your Results
The Idea: I've learned this one the hard way.
Sometimes I'll get frustrated because an AI tool doesn't understand what I want. But when I look back at my instructions, I realize I left room for interpretation.
Joe compared working with AI to managing a new employee. You can't give vague directions and expect consistently great execution. Clear objectives, specific context, and direct feedback make a difference.
Your Move: The next time AI gives you a disappointing result, don't immediately switch tools. Rewrite your instructions to include the objective, relevant background, constraints, and what a successful outcome should look like.
04AI Costs Can Get Out of Control Faster Than You Think
The Idea: One of the things I've noticed while building with AI is how quickly the costs can add up.
You start with one agent, and before you know it, that agent has called on a bunch of other agents to help finish the job. Suddenly, you're wondering whether hiring another employee would have been cheaper.
Joe explained that companies are working toward smarter ways of choosing different AI models based on the complexity of a task, rather than using the most expensive model for everything.
Your Move: Review what you're spending on AI tools, subscriptions, and usage. Identify which workflows deliver measurable value and which ones are consuming resources without meaningful results. Don't just measure how much work AI completes. Measure what that work is worth.
05Your Company's Knowledge Could Become Its Biggest AI Advantage
The Idea: Cisco has spent decades learning how networks operate, fail, and get repaired. That experience creates a body of knowledge that competitors can't simply reproduce overnight.
Joe explained how Cisco is using its accumulated expertise to build more specialized AI capabilities.
It got me thinking about how much valuable knowledge exists inside businesses that never gets organized or used to its full potential.
Your Move: Identify information your company has accumulated that would be difficult for competitors to replicate. It might be customer insights, operational history, internal processes, or industry-specific expertise. Start organizing that knowledge so your team can use it more effectively, including with AI where appropriate and secure.
06AI Should Free Your Best People to Do Their Best Work
The Idea: The biggest opportunity isn't simply getting employees to finish the same tasks faster.
Joe talked about how AI can reduce the constant firefighting that keeps talented people from working on the problems that actually require their judgment and experience.
That's a different way of thinking about productivity. Instead of asking how much work you can automate, ask what your people could accomplish if they weren't buried in repetitive tasks.
Your Move: Ask your team which recurring tasks consume the most time without requiring much judgment. Pick one to simplify or automate, then decide how that recovered time can be invested in higher-value work.
Connect with Joe Vaccaro
Joe Vaccaro is the SVP of Network Platform and Assurance at Cisco, where he works on technology that helps businesses understand and improve the networks and connected experiences they rely on.
Before joining Cisco, Joe was VP of Product at ThousandEyes, a technology company that was subsequently acquired by Cisco.
His work focuses on network intelligence, AI-powered operations, and helping organizations adopt emerging technology in practical, responsible ways.
🌐 Cisco: https://www.cisco.com
💼 LinkedIn: Search for Joe Vaccaro at Cisco.
So, What's Next?
I've worked with some of the biggest companies in the world, and one thing I've seen firsthand is how slowly large organizations can move. There are good reasons for that. When you've got customers, employees, systems, and a reputation to protect, you can't just throw caution out the window.
But what Joe reinforced for me is that AI is changing the amount of time businesses have to make decisions. Things that once took months to explore can now be tested much faster. That doesn't mean you abandon good judgment. It means you need to get better at making informed decisions quickly.
And you can't do that if you have no firsthand experience with the technology.
You don't have to become a software engineer. You don't need to understand every model or chase every new AI release. But if you're responsible for leading a business, you should understand what these tools are capable of and where they can create meaningful value.
The biggest risk isn't that AI will make every decision for you. It's that you'll be making decisions about AI without understanding it.
So open the tools. Build something. Break something. Learn what works.
Then figure out how to make it work for your business.
See you next week.
Cheers,
Ryan Alford
Host | Right About Now
CEO | The Rad Collective LLC
