We talk about AI constantly but most small teams are still using it wrong — what's actually working for you?

I've been going deeper into how AI is actually being used across different business functions and there's a pattern worth talking about.

Most founders I see either go all-in and try to automate everything at once, or they completely ignore it because it feels overwhelming. Both approaches tend to fail.

What actually seems to work is picking one painful, repetitive problem and solving just that first. Lead qualification. Invoice processing. Ticket routing. Not "AI strategy." Just one workflow.

A few things that stood out to me from real business cases worth discussing:

Finance teams using AI for expense tagging and fraud detection are cutting false positives by 40-60% compared to traditional methods. That's not a small gain.

Customer service teams handling Tier-1 questions through AI chatbots are reporting a 37% drop in first response times. Again, not replacing the team,just removing the repetitive load.

The ROI question is real though. Around 75% of companies have adopted AI in some form, but only about 25% are actually seeing returns right now. That gap is mostly because teams skip the pilot phase and go straight to full implementation.

The build vs buy debate also comes up a lot. Off-the-shelf tools get you moving fast. Custom builds give you control but cost more upfront. A hybrid approach is start with existing tools, build custom once ROI is proven, seems to be the most practical path for smaller teams.

Curious what's actually working in your business right now. Are you seeing real returns or still figuring out where AI fits?