Insights

Switzerland's regulator says the question is no longer whether banks use AI, but whether they control it. A real AI strategy is a map: where AI decides, where people stay in the loop, and which projects to stop.

Law firms, legal-tech vendors and in-house teams are racing to hire legal engineers. The title is new. The work isn't. Without a map of which legal work AI should run, check or only inform, they'll automate the easy parts and blur who is accountable.

Swiss private banks grew assets last year and still got less efficient. Most run five transformation programs at once: operating model, data, AI, technology and skills. The order matters more than the budget.

Over 40% of agentic AI projects will be canceled by the end of 2027. Recording how work gets done isn't the fix. Someone has to decide how it should get done.

A company that sells AI agent infrastructure just published data proving the infrastructure isn't the bottleneck. Here's what actually separates the companies getting AI into production from the ones stuck in pilot.

Most AI Skill pitches fail before they start, and nobody checks for it. Here's the one question that disqualifies half your list before anyone spends a dollar scoping it.

Everyone wants AI in the front office first. The readiness data says start in the back. Here's the tier map that shows why.

80% of enterprises now run AI agents in production. Confidence in securing them averages 2.32 out of 5. That gap isn't a security problem, it's an operating model that never assigned ownership.

Most business cases don't fail because the investment was wrong, they fail because the number behind them can't survive one good question from finance. This piece breaks down the method Optivance uses to build ROI numbers that hold up under real scrutiny, not just in a slide.

Capital markets are starting to question AI capex ROI. When cheap money funded AI pilots, few asked if they moved the P&L. When scrutiny returns, the gap between automated dysfunction and redesigned operating models becomes visible, and expensive.

Full-stack AI vendors offer speed and simplicity, the same pitch as a car lease. For a regulated institution, the exit clause matters more than the monthly payment.

A real IT Project Manager posting at a major international NGO lists "define the operating model" as responsibility #9 of 9, after mobile app architecture, defect resolution, and release planning. That placement is the diagnosis, not a footnote.

McKinsey's new AI-scaling data shows 89% of enterprises are still running AI on Industrial-era hierarchy. The drop-off to Agentic isn't a tooling problem, it's an org-chart problem.

The FSB's AI accountability consultation closes this week — and exposes a question most regulatory functions can't answer: who is this process actually for?

The EU AI Act didn't create your operating model problem. It just made it visible.

Every large institution has a process rebuilt five times: once per system, once per team, because nobody was ever assigned to own it end to end. Layering technology on top doesn't fix that. It just automates the duplication faster.

ExCo approved the launch. The business case cleared the hurdle rate. Six months later the product is late, over budget, and under-earning. The cause wasn't execution — it was six operating-model assumptions nobody tested before saying yes.

Development-finance institutions are being asked to find new trading partners and commercial revenue as donor support contracts. Most of them are reaching for the same playbook that worked when capital was abundant. Here's why that playbook will fail, and the strategic discipline that replaces it.

The "low adoption" of your AI pilot isn't a change-management problem. It's a rational response from people who can read the business case as well as you can. Here's the operating-model fix.

Every transformation project starts with a process map. Most of them describe work that doesn't actually happen, and the gap between the map and reality is where the customer-value question gets buried, and where AI ROI quietly dies.

There's a single question that tells you whether an AI pilot will deliver, and most teams can't answer it. Here's the test, and what it reveals about the project before a single line of code is written.

A generative AI pilot can hit every technical milestone and still die. Here's what that failure actually looks like from the inside; and why the cause is almost never the model.

MIT's State of AI in Business 2025 report found that 95% of generative AI pilots inside large enterprises deliver no measurable P&L impact. The models work. The operating model around them doesn't.