What we hear in the room

Every month we write up the patterns we keep noticing in the AI workshops we run for small business teams. Where people get stuck, what they push back on, what's clicking, and what's shifting. No specific clients, just trends.

01The agent vs. chatbot confusion

More than half the workshops this month opened with the same mix-up: people thinking "agent" just means a fancier chatbot. We've started spending the first ten minutes on the difference, because every conversation downstream gets muddled if we don't.

The short version we land on: a chatbot answers when you ask. An agent does things on your behalf, including things you didn't specifically ask for. Different risks, different setup, different kinds of mistakes to plan for. When teams understand the difference, the rest of the workshop goes smoother.

02The most common failure point: vague prompts

We watch a lot of people type things like "write me a marketing plan" into ChatGPT, get a generic answer, and conclude AI doesn't work. They're not wrong about the answer. They're wrong about the diagnosis.

The fix isn't a better tool. It's spending two minutes up front telling the AI what business they're in, who the customer is, and what they've already tried. When people see the difference that makes, it's usually the moment something clicks for them about how AI actually works.

03The resistance we hear most

Three pushbacks come up almost every workshop, in this order:

• "It sounds nothing like us." Fair. The default voice of every AI tool is bland by design. This is a context problem, not an AI problem, and it's fixable.
• "I don't want to seem lazy." This one's cultural, not technical. Teams that use AI well treat it like using a calculator, not like cheating. Permission from leadership changes the room.
• "What about privacy?" A real concern, with a real answer that depends on the tool and the data. Worth its own session, honestly.

04A small thing that's clicking

A simple workshop exercise that keeps landing: have everyone open the same prompt in ChatGPT and Claude side by side, then compare the answers. People who've been using one tool for a year often realize they have strong opinions about the other in about ninety seconds.

The point isn't to pick a winner. It's that AI tools have distinct personalities, and matching the personality to the job matters more than people expect.

05What we're paying attention to

A few trends we're tracking heading into next month:

• Teams using AI to skip the "first draft" phase across the board: proposals, briefs, replies, reports. Almost nobody is still hand-writing first drafts.
• A new wave of curiosity about MCPs: what they are, whether they matter for small teams, and whether it's worth waiting before wiring tools together.
• Growing comfort with letting AI handle internal-facing tasks, and continued (warranted) caution about customer-facing ones.

If any of this sounds familiar from your own team, send us a note. We'll tell you honestly whether a workshop would help.