She Replaced Live Demos with a Pre-Recorded Video
A live demo froze in front of a client. After that day, she only played videos recorded the day before: every step exact, nothing could go wrong again.
“Look how AI native Anthropic is, shipping a new feature every day.” Behind that line was a young, inexperienced team being dragged into a sprint it couldn’t sustain.
Later she joined an AI company. There, odds are she will once again open a product that might freeze, in front of people, instead of a video.
A live demo, and the product froze right in front of the client. After that day, Lilly never again gave a live demonstration of her team’s product to a client.
She switched to videos. On the screen was a walkthrough recorded the day before: every step exact, no chance of another failure.
Lilly worked in a newly formed AI department inside a traditional enterprise, in charge of R&D. The team had been pieced together from within the company: a few data analysts, one data engineer, one full-stack engineer. She was the product manager, running the team alongside a tech lead. Above them sat a department head whose background was strategy. The team was young and inexperienced, and what they built was never quite stable.
As a product manager, Lilly’s daily workflow went like this: first finish the PRD, then walk it past every stakeholder one by one for alignment and sign-off, so that no one could disown it if things went wrong later; then break the PRD into features and write a detailed user story for each one, in text, images, or video, whatever made it legible to the engineers; then go through those features with the tech lead to check which ones were actually feasible and which needed to be split finer; and finally pull the whole engineering team into a meeting to lock in the implementation plan and the division of work. Sign-off, breakdown, alignment, assignment. Not one step could be skipped.
What Lilly could not stand was a line the tech lead kept repeating.
“Look how AI native Anthropic is, shipping a new feature every day. We should iterate at that speed too.”
That manufactured urgency wore the team down. Rushing made what they built even less stable. Most of the team were new hires who didn’t dare push back. Only Lilly objected.
“Anthropic has more people, more engineers, a powerful R&D organization, and money it can’t burn through. Of course it can ship a feature a day. Every feature we build has to bring the company business value.”
Later, the department head from strategy left the company. The new head came from product. A while after that, Lilly left too. She joined an AI company, still as a product manager.
She didn’t come away hating demos. At the new company, odds are she will once again be in some meeting room, opening a product that might freeze in front of everyone, nervously waiting for it to load. Instead of opening a pre-recorded video.