AI strategy shifts from models to deployments

The latest research from Menlo Ventures confirms what we’re seeing firsthand with clients: companies have moved well beyond testing AI – they’re now using it to run their key business operations. Spending on AI services has more than doubled to US$8.4 billion in just six months, with three-quarters of companies now deploying AI systems that directly serve their customers and operations.

Two findings mirror what we’re seeing with clients. First, performance consistently trumps price –enterprises are willing to pay premium rates for frontier models that deliver superior results.

We see this daily as clients prioritise capability over cost savings when selecting their AI platforms.

Second, the focus has shifted decisively towards agentic applications. The most successful implementations we are supporting are not just automating individual tasks but creating AI systems that can reason, use tools, and work autonomously across complex workflows.

What does this mean for your AI strategy? Model performance remains the critical selection criterion, but the real competitive advantage lies in identifying the right use cases where agents can transform entire processes.

As you scale these applications, that’s when cost optimisation becomes crucial. The organisations that master this balance – extracting maximum value whilst managing costs effectively – will have the financial flexibility to invest in additional AI capabilities whilst their competitors are constrained by budget overruns from inefficient implementations. This isn’t just about cutting costs, it’s about allocating your resources strategically and unlocking budget flexibility to continue innovating.

Whether you’re starting your AI journey, building your first production systems, or scaling existing implementations, a pattern is emerging: companies deploying agentic applications are achieving superior results compared to those automating individual tasks. The real divide isn’t about AI adoption – it’s about how quickly you can make it work effectively.