Last Tuesday, I watched a founder pace across her screen. She'd just told me her team had automated 30% of their weekly reporting using internal AI tools. "I feel guilty," she said. "Like I'm cheating somehow." This conversation—this feeling of unease about doing AI "the right way"—has played out in different ways across all 12 companies I've worked with over the past 18 months.
What started as a straightforward project—help small businesses build internal AI capabilities—became something more nuanced. The companies ranged from a 35-person logistics firm in the Pearl River Delta to a 50-person marketing agency in Chicago. Their challenges were unique, but patterns emerged that surprised me.
Lesson one: The biggest bottleneck isn't technical knowledge—it's psychological. Many founders assume they need to hire AI experts or train everyone in machine learning. What they actually need is permission to experiment. At the logistics firm, we started with a simple exercise: "What's one task you do weekly that feels repetitive?" The answer became their first AI project. The technical implementation took three days. The permission to think differently took three weeks.
Lesson two: Your existing processes are more valuable than you think. We often rush to automate without documenting what we're automating. A 40-person e-commerce company in Singapore tried to implement AI customer support before documenting their current support workflow. The AI outputs were inconsistent because the underlying process wasn't clear. Once they mapped their existing conversations—how they escalated, what information they collected—the AI suddenly made sense. The tech was ready. The process wasn't.
Lesson three: The most valuable AI applications are boring. Every founder wants to talk about AI strategy or competitive advantage. The real wins come from small, practical improvements. A 35-person design firm in Portland used AI to standardize their client onboarding process. Nothing revolutionary. Just consistency. They reduced the time spent on administrative tasks by about 20% each week. That's not a (amount withheld). That's a sustainable improvement that compounds.
The pattern across all these companies is the same: AI adoption succeeds when it serves human work, not replaces it. The best implementations make people's expertise more valuable, not less. They handle the routine so humans can focus on what they do best: judgment, creativity, and nuanced decision-making.
If you're thinking about AI for your team, don't start with technology. Start with questions: What work makes your team feel stretched thin? What tasks happen repeatedly? Where does valuable knowledge get lost when people leave? The answers will point you toward meaningful applications.
The most practical takeaway? Schedule a 90-minute "process mapping" session with your team this week. Not about AI. About your current work. Document the steps, decisions, and handoffs in one critical workflow. You'll likely discover opportunities where AI can support—not replace—what you already do well. 📊
This piece is from our notes on helping SMBs (10-100 people) build their first in-house AI teams. If your team is exploring this — quick feedback and questions welcome in the comments.
Top comments (0)