3 ms·
First you need really good infra to make it easy to test working multiple solutions for AI but I think this will be bleeding edge in 2030. EDIT: with in-memory
by machiaweliczny 5y ago
First you need really good infra to make it easy to test working multiple solutions for AI but I think this will be bleeding edge in 2030.
EDIT: with in-memory DBs I can imagine AI assisted mainframe than can solve 90% of business problems.