Developers can earn buy-in from leadership teams by clearly conveying how AI projects deliver value and align with key business objectives
Despite being a clear focus for companies across a wide array of industries, it’s expected that numerous AI or machine learning initiatives will become stuck in the pilot phase over the next 12 months, with this stagnation commonly referred to as pilot purgatory.
Gartner estimates at least 30% of generative AI (GenAI) projects will be abandoned after the proof of concept (PoC) phase by the end of 2025, primarily due to poor data quality, inadequate risk controls, escalating costs or unclear use cases.
According to Bartek Roszak, Head of AI at STX Next, development teams should consider a project’s feasibility and potential from the outset to earn the backing of the C-suite and successfully transition from PoC to production.
Roszak said: “A swathe of companies have rushed to implement GenAI solutions, but the reality is that the bulk of these projects will never come to fruition. Gartner’s prediction that 30% will be abandoned looks to be conservative at this stage – this figure is more likely to be around the 75% mark.
“For many C-suite leaders, AI is still relatively new, which can create hesitation when it comes to making considerable investments in the technology. To gain their support, it’s important to first present a clear outline depicting how AI can transform the business and satisfy long-term goals.
“After establishing the vision, the next step is to introduce low-risk pilot projects that deliver quick, measurable returns. These pilots help prove AI’s effectiveness and demonstrate that more ambitious goals are achievable, which builds confidence and trust among key individuals in the proposed strategy.
“AI projects, like any other, must ultimately deliver clear business value. When we say AI strategies should align with business goals, it simply means that before a project even begins, there must be a well-defined objective and a consistent way to measure whether or not targets have been achieved.
“The reason we emphasise this so much now is that, in the past, many companies treated AI implementation as a goal in itself. This has led to myriad pilot projects failing to reach production because it was difficult to justify significant investment in something that didn’t clearly benefit the company.”
Roszak also believes development teams must contain strong leaders, capable of working with a range of stakeholders, to deliver AI deployment.
“Leaders need collaboration and communication skills to foster effective teamwork between AI experts, subject matter experts, and other stakeholders to encourage everyone to work towards shared objectives. Agility and flexibility are also key for adjusting plans and resources as challenges arise, keeping the project on track.
“A strong technical understanding of AI enables informed decision-making and helps balance innovation with feasibility. Finally, effective change management is essential for navigating organisational shifts, securing buy-in and ensuring smooth AI adoption.”
Roszak concluded: “Successfully scaling a GenAI project requires a solid strategy and a well-structured AI roadmap that enables a seamless journey from PoC to production, with no room for PoCs that are impractical or too costly. Following these steps can win the support of the C-suite and help initiatives avoid the dreaded pilot purgatory phase.”
