Jacqueline Cheong (Artie) built the feature all her customers rejected, and today 40% use it

Note up front: what was downloaded is not a full article but the page for a podcast episode ("Agents at Work", by Julia Fedorin, in collaboration with Composio) with a brief introduction and a chapter index. There is no transcript, so this summary is limited to what that description and the timestamps say. The episode runs about 54 minutes and was published on October 19, 2026.

The guest is Jacqueline Cheong, co-founder and CEO of Artie. According to the description, Artie moves data between databases in real time. It is the "unglamorous plumbing" that lets AI agents reason about what is happening now rather than what happened yesterday.

The central anecdote is about the "history table". According to the text, every customer told Cheong they didn't want that feature. She and her co-founder had conviction and also wanted to use it themselves, so they launched it against that feedback. At first nobody used it. Six months later, 40% of customers had turned it on, and the introduction says it is now something that 40% depends on.

Other topics the description announces:

- Founding a company with your spouse. The CTO and co-founder, Robin, is her husband. Cheong tells every new hire and every investor about it from the start. - Radical transparency. The whole team sees revenue, burn and investor updates. According to her, having full context makes every conversation easier. - Shifting focus from the final metric to the "inputs", that is, the actions that produce it. - Why real-time data is the layer AI agents truly need.

The chapter index includes: what Artie does, why real time is hard, the AI agent angle, the history table, founding with her husband, transparency with the team, a "Working With Me" document, focusing on inputs and a wedding cake metaphor. No detail is given about the "Working With Me" document or the wedding cake metaphor beyond their titles.

Since the page only contains the description and the index, I cannot verify figures, reasoning or quotes beyond the above. For the details (how they designed the feature, why customers rejected it or how she measured its adoption), you need to listen to the episode.