Deutsche Bank’s investment-banking CIO Denis Roux recently spoke with Reuters and reported on AI project timelines: projects that once took two years now take three to six months. That is not a projection. That is an operational result from a major global bank.
What Shorter AI Project Timelines Mean in Practice
First, speed at this scale is structural, not cosmetic. Two-year projects become six-month projects because the cost of iteration collapses. Testing assumptions, revising requirements, rebuilding after learning more. All of it moves faster. That does not just save calendar time. It changes how organizations plan, budget, and compete.
Second, Deutsche Bank is running a deliberate cost discipline alongside the productivity gains. Engineers receive token budgets. They can request more, but only after demonstrating value. Learnings are then shared across the organization. That is exactly the discipline companies developed when they moved to cloud computing, and it is the right instinct. Usage-based AI pricing is where the market is heading. Anthropic and OpenAI have already been shifting enterprise customers to token-based models. Deutsche Bank is building the internal governance to match that reality before it is forced to.
Third, the use cases driving these results are worth noting: financial data analysis and portfolio exposure checks. These are repetitive, high-stakes analytical tasks. Not because the work is trivial, but because it is structured, rules-bound, and verifiable. Human judgment stays on the decision. The AI handles the data assembly and analysis cycle underneath it.
Most enterprise AI conversations are still happening at the strategy level. Deutsche Bank is running experiments, measuring outcomes, and scaling what works. That is a different conversation. And it is the one that matters.
Source: Reuters, June 18, 2026
Get new articles from Robin Green delivered directly.
Insights on AI leadership, the future of work, and human collaboration. No noise. Unsubscribe any time.
Subscribe free