ONE. https://computer.rip/2025-06-08-Omnimax.html - the history of Omnimax (one form of IMAX).
The TLDR - never profitable; it's amazing a project kept afloat by "science museums" for decades did as well as it did.
TWO. - LLMs are Cheap ⚙️ https://www.snellman.net/blog/archive/2025-06-02-llms-are-cheap/
It may be time to start a "LLM-powered widgets" business.
The word widgets is a bit of legerdemain; it distracts from the details. the process would be:
* choose a vertical (education, "personal productivity", law, medicine, etc.)
* get a team of about five (one designer, two developers, one quality-assurance, and me)
* build a suite of tools (for "education", not just "a reading game to learn -op", but "reading games for everything in the 1st grade curriculum) ⚔️ for "law", my cheapest estimate includes $500k to train models on specific corpuses of government legal codes and court decisions. for education and "personal productivity", the regular APIs are enough.
* get to an MVP within 3 months
* sell sell sell
There's a new soccer tournament ⚙️ the Club World Cup, and it probably will suck.
but: it is a dry-run for the World Cup next year. and, for "does Trump blow up border-control to the point this doesn't work". ⚙️ we invoke Guild Law: no discussion of politics or religion