The best person at AI in your whole company might be the one who, a month ago, had never even heard of it.
I’m Tim, COO at Bridge AI Consulting, stepping in for Dean this week, and I’ll be honest, it surprised me when I first saw it in practice. Anthropic went looking for the people who get the most out of Claude, and the answer wasn’t the youngest folks or the engineers. It was the people who understand their own job and how they do it better than anyone, whether they’ve written a line of code in their life or not.
Some people read that and think no way, the technical people have to be winning. Others read it and think of course, the person who knows the work knows what good looks like. Both reactions are fair, and both are staring at the exact same technology. That gap is what I want to get into today.
This week: the surprising thing that separates people who get real value from AI from everyone else, the agent that turns a startup’s week of lead-hunting into a list waiting in the inbox by Monday, and three stories from a loud week in AI (a Chinese open model sneaking up on the giants, SpaceX casually dropping $60 billion on an AI startup, and a survey that says the more we use AI, the less we trust it).
THE DEEP DIVE
Three Weeks Ago, He Asked “What Is AI?” Now He’s the Best on His Team.
Three or four weeks ago, a project manager in his fifties at one of our clients heard a colleague bring up AI and asked, dead serious, “What is AI?” He had never touched it. Three weeks later, he was the one building the most with AI on his entire team.
He’s not a fluke. Anthropic just went through 400,000 sessions in Claude Code to see what drives results, and it wasn’t coding ability. It was how well each person understood the problem they were solving. In their words: “success is determined by how well a person understands the problem they are trying to solve, not whether they’re trained in coding.”
And this was a coding study. If technical skill isn’t what wins even at writing code, it isn’t what wins anywhere else you put Claude to work either: your emails, your client decks, your reports, your research. The lesson isn’t about coding. It’s about how you get the most out of Claude.
Their own example says it best. They describe “an accountant who has never used Python, but tells Claude exactly which reconciliation rules a Python script must enforce and catches the edge case it mishandles at month-end close,” and call that person an expert at the task. Knowing the work beat knowing the code, and that was the line between the people getting real value and the people getting a shrug.
We see it on every team we train. The people who save the most time and build the best things are almost never the most tech-savvy. You don’t need to learn a coding language or anything technical. You need to understand your own workflow well enough to explain it.
The mental model I give people: talk to Claude like a sharp assistant you hired yesterday. Ask someone brand new a point-blank question about something they’ve never seen and they’ll guess, and the guess won’t be great. Even with the entire internet at their fingertips, they still wouldn’t know how your business runs, where your numbers live, or what your customers care about. The way you turn that new hire into someone who saves you hours is you walk them through it: what you look at, where you pull the data from, the edge cases, a few good and bad examples, the things you specifically check for, and the catches that aren’t obvious until they bite you. Then you show them what a great finished product looks like, the one the client wants to see. Do that, and Claude works at your level, because you handed it what was in your head.
That’s the first step: understand your workflow well enough to describe it, hand it off, and check the result. The bigger unlock comes from those same people. Because they understand their work and their department better than anyone, they’re the ones who can step back and see it from a bird’s eye view. Now that Claude can analyze a mountain of data in minutes, or turn out a client deck before lunch, they ask the better question: what can we finally do that we never had the time for? What would wow this client? That question was always sitting there. There was just never time to chase it. Now there is.
This is the part of AI you can’t buy off a shelf, and it’s part of what we do with a team. We don’t just automate the workflow. We teach your people to direct Claude like the experts they already are.
FROM THE FIELD
50 Leads in the Inbox Every Week
Finding 50 genuinely good leads a week, by hand, is the kind of job that eats a person’s entire week: scrolling player profiles one at a time, cross-checking stats and team pages, then writing to each one. NXTPlayFB is an early startup, and like every business, it runs on reaching the right people. It helps football players manage their whole career, from their first recruiting email through college and into life after the game. What it didn’t have was anyone with a spare week to go prospecting.
So we built them a lead-generation agent. Every week it crawls social media and team and stats pages, pulls candidates into one system, and scores each one against the owner’s ideal-client profile, which he can adjust as he learns what converts. It hands back 50 ready-to-go leads a week, each with a personalized outreach message drafted and ready to send, and it runs A/B/C tests on the messaging to learn what gets players to respond.
What used to be hours of manual sourcing every week is now a list of 50 scored, personalized leads the owner acts on with a click. And it compounds. Every result feeds back into the scoring and the messaging, so the list gets sharper over time. A human still reviews and sends every message.
Here’s why this travels: lead generation is the one job every business shares. A contractor, a law firm, an accountant, a SaaS startup, they all live or die by finding the right people to talk to. The same kind of agent works for any of them, dropping a batch of researched, scored, ready-to-contact leads in your inbox every week.
AI NEWS WORTH YOUR TIME
Three Stories From This Week
A Chinese open model is catching up to the big names
Z.ai, a Chinese lab, released GLM-5.2, an open model that lands close to the top frontier models on coding benchmarks. It’s the best open model anyone has put out, it runs at a fraction of the cost, and the weights are free to download under an open license. The coverage this week was less about the model itself and more about what it signals: cheaper, open models putting real pricing pressure on the expensive ones.
My take: open source models are no longer far behind the frontier, and that’s worth noting. GLM-5.2 still isn’t as good as Claude or ChatGPT today, and those two are far better and far easier to use, so I wouldn’t expect anyone to switch right now, and I’m not telling you to. But keep an eye on it. If the premium models keep raising their prices per token, a cheaper open model you can run and customize with your own proprietary data could become a real option for some businesses down the road. Not today. Just a door worth watching.
SpaceX bought an AI startup for $60 billion
Days after the biggest IPO in history, SpaceX bought Cursor, a popular AI coding tool, in a $60 billion all-stock deal. Elon Musk became the world’s first trillionaire in the same stretch. A tool that many developers rely on got swallowed into someone’s empire basically overnight.
My take: this is bigger than a tool changing hands. Cursor brings a massive base of working software engineers, and SpaceX brings serious computing power. Put those two together and you could accelerate how fast AI development moves. The broader signal for your business: when the tools you rely on keep getting absorbed into giants, stay flexible. Build your workflows around your workflows, so when something better or cheaper comes along, you can swap the model underneath without starting over.
Half the country uses AI now, and trusts it less than ever
New Pew data this month shows about half of US adults now use AI chatbots, up from a third in 2024, and roughly a quarter use one every day. But the more people use AI, the less they trust it. 63% say it’s moving too fast, and 71% think it makes their personal information less secure. The heaviest users are the most skeptical of all.
My take: adoption isn’t the question anymore, trust is. Your customers and your team are already using AI, and they’re wary of it at the same time. For a business, that means being upfront when AI is part of how you work, and keeping a person accountable for anything that reaches a customer. The companies that are honest about where the human still sits will earn the trust this data says is running thin.
That’s it for this week.
If this was useful, forward it to another owner who’s trying to figure out where AI fits in their business. That’s who this is for.
And if you want to talk about what AI could do for your business, not in theory, but in your workflows, with your team, on your timeline, reach out at [email protected]. That’s what we’re here for.
Tim built the internal AI workflows and training systems that got developers at Epic Systems using AI every day. He's a Partner and COO at Bridge AI.
