AI SDLC transformation
Development gets AI last. By putting AI into planning and feature prep first - on top of DDD and docs-as-code - delivery reached ~2x greenfield and +60% brownfield, with overtime and incidents nearly gone.
Read the caseThree cases, one through-line. I was fast before AI (a working end-to-end banking PoC in 10 weeks, no AI), I rebuilt stalled processes into a healthy foundation, and because of that foundation my AI rollout worked where the naïve version fails.
Development gets AI last. By putting AI into planning and feature prep first - on top of DDD and docs-as-code - delivery reached ~2x greenfield and +60% brownfield, with overtime and incidents nearly gone.
Two replacement projects that had produced nothing - 2.5 years on anti-fraud, 13 months on processing - turned into steady, low-rework delivery through Kanban, DDD, docs-as-code and ADR/RFC.
A team built from scratch shipped a working end-to-end digital-banking PoC in 10 weeks - no AI - straight in the target architecture. The Minimum Lovable Product is now in the central-bank audit.