The best AI isn't impressive demos — it's time saved before a high-stakes conversation. These agents solve the problems I've watched executives hit repeatedly. Each one is built to produce something you can act on, not just read.
Enter who you're meeting and what you want to achieve. The agent synthesises everything it knows about them and returns a one-page brief — ready to download, print, and walk in with.
Under the hood
A note on how this is built
This version runs on Claude's training data alone — no live web search. That means it works best for well-known executives, major Australian organisations, and publicly documented strategic moves. It won't know about news from the last few months or people with a limited public profile.
A live web search version exists and produces richer, more current briefings — but at roughly 50–100× the token cost per call, making it impractical to offer freely. This is a deliberate tradeoff: a tool that works reliably for most use cases, at a cost that doesn't require a subscription to run.
Why I built this
Fifteen years of walking into high-stakes meetings with large organisations taught me that preparation is the only variable you fully control. Everyone else in the room has done their research. The question is whether yours goes deeper.
The problem isn't access to information — it's synthesis under time pressure. A CSO preparing for a partnership meeting doesn't need a Wikipedia summary. They need to know what that person cares about strategically, what their company is betting on, and what questions will signal genuine preparation rather than a Google search.
I also built this to understand the real cost and complexity of agentic AI tools in practice. Adding live web search multiplied the cost per call by 50–100× — a genuine product decision, not just a technical one. Knowing where to draw that line is part of building AI products that actually scale.