W19 - NotebookLM, DeepWiki and AI Products

Recently I’ve been using information-organization tools like NotebookLM and DeepWiki. NotebookLM is designed around the concept of a workbench and has an IDE-like feel. Moving this to an “at work” scenario, a workbench that can freely integrate internal content sources such as dx, ones, code, and km could multiply productivity. Having studied the OODA loop, I see AI as accelerating its cadence, especially in the Observe and Orient stages. It helps us extract and understand key information faster and build cognitive frameworks, enabling more frequent strategic adjustments and decisions. For work like writing a COE, the reviewer only needs to focus on root causes and lessons learned—the timeline and similar details would be filled in automatically, and anyone could dig into an incident at any time; that’s what a true case study looks like.

I recommend an episode of Lenny’s Podcast that interviews OpenAI’s CPO—an excellent product-perspective deep dive. If you want to save time, use NotebookLM. Three points left a strong impression on me.

1. The tech stack that products rely on is shifting. We’ve moved from rule-based systems of “fixed input—fixed output” to model-based systems that produce “probabilistic outputs.” Models iterate extremely quickly; every couple of months computers can do things they couldn’t before. The model you use today will almost certainly be the worst one you’ll have used in the future, so you must continuously adjust product thinking and pacing.

2. Evals are becoming a core competency for product managers. As Yao Shunyu described in the “LLM second half,” it’s not just about testing models but about evaluating the performance of the whole technical system on specific problems. Products need to build models tailored to real problems rather than waiting for a model to be “perfectly ready.”

3. AI won’t generate the ultimate creative idea with a single click; it accelerates the process of bringing ideas to life. It can quickly provide multiple options for creators to experiment with, combine, and refine. The real productivity gains come from this iterative “human and AI co-creation” cycle.