The Drip
AI activity keeps climbing while public sentiment keeps getting worse. Pete Halupka sits right in the middle of that tension: he works in renewables, spends time with people trying to use AI in sales, and is skeptical enough to ask what all of this infrastructure is actually for. The useful question is whether we can form an informed opinion by using it and stay honest about the tradeoffs.
Inside The Bottle
A banner over a Birmingham highway said, “AI is for dummies.”
Pete said that a younger version of him might have dropped it.
Instead, he started using the tools. He built a HubSpot workflow that removed a painful CSV import. He made an ElevenLabs cold-call coach so sales reps could hear objections and practice answering them out loud. The point was to remove toil and create more room for higher-quality connection.
That led us to Toyota.
In a highly automated system, the dangerous move is pushing every quality check to the end. Toyota puts quality gates throughout the line, with people able to stop it when something is wrong. AI needs the same treatment. Human judgment should not be a final approval step bolted onto an autonomous process. It belongs throughout the work.
The same idea applies at the company level. You do not show up with a separate AI strategy and look for places to deploy it. The strategy is already hiding inside the company's existing priorities: growth, stability, customer experience, the work people are actually trying to get done. The job is to discover where AI fits those priorities and where it does not.
That discovery starts with people. Some are excited. Some are terrified. Most are busy doing the job they already have and do not spend all day drinking from the AI fire hose. A strategy has a better chance when the people doing the work can see the destination and help choose the route.
Curiosity is not blind enthusiasm. It is a willingness to use the technology, notice what it changes, and keep enough skepticism in the process.
Lab Notes
| ■ | Justin's note: AI may be the most amazing technology of our lifetime and the most challenging implementation of our lifetime. Both can be true. |
| ■ | Kellan's note: If the quality gate is only at the end, the system was not designed with enough human judgment in it. |
What Stopped Our Scroll
| ■ | Don't Paste the AI — A sharp reminder that people want your judgment, not a model answer pasted back at them. |
|
