If every AI headline feels like it’s written for Silicon Valley, this one’s for you. I lead marketing for a business built on trucks, technicians, and locations, about as far from a tech startup as it gets. Here’s how a normal company actually starts.
Start with three tasks, not a strategy
Forget the big transformation plan. Pick three recurring tasks that are annoying and low-risk, and try AI on them for two weeks:
- Writing that nobody enjoys. Job postings, policy updates, customer replies, internal announcements. AI drafts, a human approves.
- Summarizing long things. Turn a 20-page report or a messy meeting transcript into five bullet points anyone can act on.
- Answering the same question over and over. Draft a friendly, on-brand answer to your ten most common customer questions, once, then reuse.
Give it a week, then look at the time
The only metric that matters at the start is hours returned. If a task that took most of a morning now takes a few minutes, that’s your proof. Real companies fund what saves time, not what sounds futuristic.
What to ignore (for now)
- Custom-built AI systems and big consulting projects. Too early.
- Anything that touches sensitive customer or employee data before you’ve checked your policies.
- The pressure to “have an AI strategy.” You earn the right to a strategy by first proving value on small, real work.
What actually separates the winners
The companies pulling ahead aren’t the most technical. They’re the ones who started small, paid attention to what worked, and kept going. You can be one of them by Friday.