Archive

Today's signal

The companies getting value from AI are redesigning how work is done, not just deploying tools, and that is now the technology leader's job.

In the enterprise

Genpact: 79% of enterprises have AI agents in production, but only 15% tie AI spend to profit and loss

Source Genpact

What happened

Genpact surveyed more than 2,000 executives. Only 20% use AI agents widely, only 49% of funded AI projects delivered the value expected, and only 9% have significantly changed how they're organised. Also, 92% of companies using agents kept their workforce the same size or grew it.

So what

Deployment is no longer the bottleneck; the operating model is. Boards will want to hear how AI spend is tied to profit, and whether the organisation has actually changed how it works.

Talent & organisation

Canva cuts about 1,300 roles to rebuild itself, and its founders say they'd step aside if they were no longer the right people to lead

Source Forbes Australia

What happened

Melanie Perkins said building products had started to need "more handovers and coordination than necessary." The cuts come as Canva faces pressure from ChatGPT and Claude, and some investors have marked its valuation down to as low as US$30.4b.

So what

With AI, the time teams spend coordinating is now a strategic cost. Technology leaders should be ready to explain how they would cut handovers without hurting reliability.

Cost, cloud & markets

Xero lifts guidance, starts a A$1b buyback, and launches its "Xforce" engineers

Source Forbes Australia

What happened

Xero's half-year revenue rose 46% to about NZ$1.7b, or 27% if you don't count its Melio acquisition. It raised its full-year guidance, and the shares jumped about 9%. Xero also says 97% of its engineers use an AI tool every day, and its CFO, Claire Bramley, leaves on 31 January. Separately, it launched Xforce, which sends engineers to work with accounting firms and rebuild how they work on Xero's AI platform.

So what

Xero is betting that AI pays off when you redesign how customers work, not just when you ship features. Forward-deployed engineers sitting with customers are a practical way to turn a platform into a change in how work gets done.

The idea

Why even your best employees override AI systems that work

Source Harvard Business Review

What happened

Das Narayandas argues that many AI systems stay stuck in an advisory role because companies give the AI the decision but leave the blame with people. When someone is still personally on the hook for the outcome, overriding the AI is a sensible way to protect themselves.

So what

Adopting AI is a leadership problem, not only a technical one. If people still carry the blame when an AI decision goes wrong, they will keep overriding systems that work.