How quickly can you get a production AI system running?

That depends on scope. Most engagements deliver a working increment within weeks. The Velocity Library compresses delivery by drawing on patterns from previous engagements rather than starting from scratch.