Why deep research got worse, and how to get it back

Here's something nobody is telling you: deep research has gotten worse over the last six months, and you're paying the same for it. The models didn't get dumber. More of the web started blocking or charging the agents your AI sends out, so every report you run is built on a shrinking slice of what's actually out there. If your results have felt thinner lately, that's why.

In this session I'll show you how I get consultant-grade research anyway. We start with the decision, not the prompt. We design the report before we search. And we let the AI interview us and write the research prompt itself, which is almost always better than the one you'd type. Then we run it live on a real go/no-go call about entering a new market, so you can see the difference it makes when the stakes are high and the deadline is real.

I'll also run the same play fast on a few smaller calls you probably face every quarter: deciding whether to raise prices, launching a new product/service or not, and answering Which customer segment should you go after next?

It works in whichever AI tool your company pays for. Bring a decision you're sitting on.