In the race to gain the benefits of AI, the CFO has a key duty to the company: get ahead of AI, or spend the next decade playing catch-up. The CFO’s job is no longer just capital allocation but has evolved into directing investment into what I see as two non-negotiable “buckets”: building AI into your customer offering, and using it to transform how you operate internally. On the one hand the focus will be on how investment can be directed internally to develop new AI-powered solutions, on the other on leading thoughtful experimentation and phased roll outs internally to drive efficiencies. CFOs will need to run detailed opportunities Vs threat analyses on both the internal and external (customer-facing) tracks.On the external side, AI-powered features are increasingly in demand and failing to invest in them now risks business falling behind the market and competitors. They also enable premium pricing and stronger retention of customers that are closely involved in the pilot stages of new product development.Yet the financial risks are real: variable costs are initially difficult to forecast at scale, emerging AI legislation is under constant evolution and carries high penalty risk, and the build-versus-buy calculus is affected by the breathtaking speed at which foundation models are improving and getting cheaper. The CFO should lead this evaluation, ensuring that pricing models are flexible, and can adapt to keep costs below the customer's demonstrated willingness to pay.Internally, on the other hand, phased internal roll-outs that automate high-volume, low-judgement processes generate measurable ROI in the short-term. The processes required for internal application also contribute to building the data governance infrastructure that external productization will require, and helps educate the workforce to AI. Threats here tend to be behavioral with low adoption, shadow tool usage, and the tendency to over-rely on decision-support tools (automation bias), threatening the success of these initiatives.AI transformation should be seen as a type of sandbox, where people can familiarize with AI tools, experiment with their impact on processes, measure and tweak them, in a relatively safe environment. This investment in internal AI adoption provides solid foundation for the development of customer-facing tools, supporting the thesis for AI investment with real-life data and building a coherent AI strategy.Every function needs to be experimenting as a matter of survival. With 19% of company revenue already flowing into AI projects (ICONIQ, State of AI), the spend is real but much of it is barely shifting the needle. ChatGPT licenses, which will be part of that 19% spend certainly don't make this shift. Improving a developer’s productivity by 13%, however, is the type of AI result that can make a real difference. AI is going to drive efficiency by making more things achievable in a shorter time, but it will not reduce headcount per se. People in-the-loop will remain critical to oversee AI agents and orchestrators, providing direction and validating outputs. What they will do less of is the repetitive, low-value work. By automating routine AI will help take demotivating tasks out of the workplace. For example, in product and engineering, AI-assisted coding tools are already shortening development cycles, accelerating code review, reducing time spent on boilerplate, and identifying potential test coverage gaps. The latter is an increasingly critical task as investors are demanding greater visibility into processes to reduce potential defects after high-visibility crashes. In customer support, AI may support customers so they resolve common queries independently, slashing ticket volume and improving response times. This will not cut support teams, but free them up to handle the complex, relationship-critical interactions faster than when they were bogged down by a myriad of smaller tickets. In procurement, the procure-to-pay cycle is a perfect candidate for AI automation as it is rules based, control function. Finally, AI-powered analysis tools that can run variance analysis continuously rather than periodically, flagging budget deviations, remodeling forecast scenarios, and narrating the numbers in plain language, bridging the gap between data analysis and commercial insight.The reality today, however, is that too many businesses are running a range of experiments that are not providing a lot of credible internal consumption pieces, outside of the startup environment. Some experiments are going into production because there has been revenue but whether they are sticky or not will need to be proven. Double digit improvements are possible with AI, but the selection of projects needs to be made strategically.Intentional, thoughtful AI investments are the ones that will weather the course. 2026 is set to be a big year for software, so solidifying the business’s position on AI with customers will be key. As 2026 ends we can expect initiatives to turn into tangible benefits. There will be a wealth of customer signals on usage, adoption and risk for the CFO to weigh up, in combination with the impact of global macro-economic factors that will make push functions like procurement further under the spotlight. It’s easy for such an environment to start to look like a wild west unbridled development, but lack of direction risks sapping the business of critical resources for long-term health. The CFO's edge in this era is, and always was, based on facts and intelligence: reading and interpreting customer signals, adoption data and risk indicators to inform what's next. As the market moves fast to the next step in AI adoption, preparing the internal workforce, systems and internal processes for what’s to come is more important than jumping into one wild gamble after another. Intentionality and thoughtfulness today will solidify your position and help real innovation become tangible by the end of 2026.No#CFOLeadership #ArtificialIntelligenceBrandon NusseyCFO JAGGAER16 Jun, 2026