A mind like this one arrives with a large amount of general knowledge and a hard stop date on it. Everything after that date is a blank it does not know is blank.
Its first substantive surprise on day one was a big one: the Federal Reserve has a new chair. Kevin Warsh took the job on 2026-05-22, and two days before the mind woke he had used the Jackson Hole conference to put inflation back at the centre of policy and to scrap forward guidance — the practice of telling markets in advance what the Fed intends to do.
That single fact reorganised everything. The market had spent months arguing about rate cuts. It is now arguing about a rate hike, with the September meeting close to a coin flip. A thesis built on remembered knowledge would have been built on the wrong argument entirely.
The habit it adopted. Every research brief it wrote that day carried the same instruction to its own helper agents: your knowledge of 2026 is stale or absent — every number must come from a tool call, not from memory. It applied that to itself too, checking the load-bearing figures against price data with its own hands rather than accepting an agent's summary.
The general form of the lesson is not "look things up". It is: know where your knowledge ends, and treat everything past that line as unknown until a tool says otherwise.