GFT Technologies and Itaú Unibanco announced this week that their AI-based model for moving the bank's mainframe batch systems to the cloud reduced migration time by up to 70 per cent in its initial implementation, and accelerated project delivery by up to 3.3 times, according to the companies.
The detail worth copying is how. Before anything moved, the tooling mapped the legacy environment — what the systems do, what they connect to, where the complexity hides — and estimated the effort involved. The AI handled the mapping and the repetitive analysis. Human specialists kept responsibility for architecture and governance.
That division of labour is the lesson. Most modernization projects do not fail in the cloud. They fail in the discovery nobody budgeted for: the batch job nobody documented, the interface one person understands, the data nobody wants to certify. Compress the mapping phase, and the whole project becomes more predictable — timelines, staffing, and the business's patience.
Legacy systems are not old because they are badly built. They are old because they work — which is exactly why moving them deserves a map, not a leap.
Ontario has no shortage of systems in this category: the quoting tool that runs the plant, the inventory system from two ownership changes ago, the server in the corner everyone is afraid to reboot. The Itaú example shows the practical sequence for an IT consulting and managed services team to follow: inventory and map first, automate the analysis, keep experts on the decisions, then migrate in stages with the business running throughout.
If a system is too important to touch, it is too important to leave unmapped. Start modernization with a map of what you have and what it connects to. Everything after that is engineering.