Notes

The demo is already in the shop. Why is the work unchanged?

2026-11-14 · Yunqing's OPC

Many stores, parks, and venues have already met AI. A speaking digital greeter may stand at the door. A chatbot may have been dropped into the official account. A report may already contain a few flattering screenshots. On the day of the demo, people nod. Two weeks later, the work that still eats the hours is unchanged: weekly posters, review replies, stock checks, tour scripts, sign-up reminders.

I call this “the demo that stays in the shop.” It looks like a start. In fact the new tool has been placed where it is easiest to show visitors. Leaders have something to look at. People in the job have not dropped a single repeating task. Before long the power cable loosens, the dialogue box sits unattended, and everyone goes back to the old way.

Demos stall less often because the model is not new enough. More often three things stack together.

First, the demo solves “it looks like we have AI,” not the action that repeats every day on the job. A digital greeter can welcome people, but greeting is often not the store manager’s real pain. The pain may be three posters in three sizes due every Wednesday, with prices that must match the system. A hall robot can answer standard questions, but a guide’s real fear is that after a new exhibit opens, the old script is still being spoken. If the pilot does not aim at that action, the demo can be complete and the business will not move.

Second, no one owns the review. When AI drafts, who is responsible for figures, tone, and permissions? If the answer is “let everyone take a look,” the draft never enters a real process. It sits in a chat, or is copied once, and the next time people go back to writing by hand. The reviewer must be the person in the role, not the person most excited about the new tool. Enthusiasm matters. It cannot replace responsibility for prices, times, and facts. A pilot without a reviewer becomes another internal demo.

Third, the knowledge is not maintained. Prices change, exhibits change, court rules change, and the demo still uses the copy from launch day. The first time it helps. The second time it is wrong. The third time no one dares to use it. Half of “landing” is agreeing who updates that small set of materials, and on what rhythm — for example, the day a price list changes, or two working days before a new exhibit opens. The set does not have to be complete on day one. It must have an owner.

So, rather than buy another platform, pick one action and write down four things: how it is done now, what done looks like, who reviews, and which sentence you will look at after four weeks. Finish that run, then decide to widen or stop. What a one-person company can help with is finishing this run — not making the demo look more like a launch event.

If you are looking at a very new demo in a shop, a hall, or beside a court, ask first whether it has touched the most time-consuming action of this week. If the answer is clear, the next spend may be worth it. If it is not, even a beautiful demo is only another exhibit that someone will have to dust.

Back to notes