Think big, start small, iterate often: A practical guide to setting up successful AI projects

Gulf Business

2 days ago

Image: Supplied
While AI and generative AI are top tech priorities for companies in 2025, with
This is often due to a lack of clear purpose, unrealistic expectations, or a fundamental misunderstanding of AI's nature. Companies tend to jump into an AI project without identifying the problem that needs solving or with the misguided notion that AI is simply a more sophisticated form of automation.
The twin pitfalls of overpromising and underdelivering are compounded when AI professionals lack the necessary project management skills to run a full-scale AI transformation project and when project managers treat AI projects like non-data projects.
The high failure rates indicate the urgency for organisations to fine-tune their strategies and align goals between key stakeholders to position their projects for success.
To drive value from AI, project professionals must identify the specific problem(s) they're trying to solve. This will help guide how they measure success – through savings in cost, time, and resources, or by improving compliance and user satisfaction.
Data is the lifeblood of any AI initiative
AI projects are not like traditional software projects. Data is the lifeblood of any AI initiative – and not just any data, but high-quality, relevant, and properly curated data. This means that project managers must take a data-centric perspective to better understand how that data drives outcomes. Using a framework that accounts for the unique demands of data-driven projects and provides clear processes and practices for data-heavy, iterative projects is essential.
To do this, project managers must learn to ask the right questions to make a critical AI Go/No-Go decision. What problem are we trying to solve? Can this be solved through other methods such as automation or programming, or does it require AI? If AI solutions are required, what is the return on investment that we are targeting? To make the project viable and ensure a sound business case, robust data availability, and supporting technologies aligned to the project needs are key requirements.
Other crucial deciding factors include establishing relevant KPIs that highlight key aspects of the project's impact; understanding how success is measured; availability of sufficient historical data; and recognizing the potential integration challenges within the organisation's existing systems.
For successful AI project management, remember this key principle to guide your thinking: 'Think big. Start small. Iterate often'.
Begin by focusing on a small but specific, well-defined problem. Look at the data available. Is it 100 per cent accessible and usable? Is it internal data? External data? How will you prep the data? Working with data you can immediately access makes it easier to iterate and move on to solving bigger challenges. In this way, the process is driven by business understanding but shaped by data understanding.
The CPMAI methodology positions AI projects for long-term success as it turns the spotlight on people and processes, making these processes more iterative and adaptable.
However, project management professionals must remember that a proof of concept alone is not enough. As AI solutions need to be trained on real-world data — rather than lab-controlled data -— building a small pilot project in the real world, using real-world data, with real-world users is what AI project leaders should aim for. This makes it easier to see how your AI solution works in the real world, allowing the team to monitor performance, gather feedback, and immediately address issues to make the model work and scale up based on the outcomes.
No room for a 'set it and forget it'
A key feature of all AI projects is that their lifecycles are continuous. Treating AI projects as one-time investments is setting them up for failure. In AI, there is no room for a 'set it and forget it' mentality as models require regular evaluation and retraining. Without this ongoing process of refinement, it can lead to misaligned ROI and poor or costly outcomes.
My colleague, Hanny Alshazly, MD for the Middle East and North Africa at PMI, has described AI as 'a game-changer' in driving the success of the region's ambitious mega projects. 47 per cent of project managers in the MENA today utilise AI to predict and mitigate risks and AI-enabled performance monitoring is shrinking project delivery timelines by 20 per cent, demonstrating its impact on overall project outcomes.
AI elevates the role of project professionals, making them strategic partners in ensuring the organisation's ability to implement AI-driven transformation. Project managers are responsible for adopting methodologies tailored to the unique challenges of AI projects such as iterative development, data-centricity, and real-world pilots. The project manager's pivotal role in a project's success makes it imperative to invest in their training and education to help organisations drive real business value from AI.
Finally, we must remember that AI is not a replacement for humans or human intelligence. It's a tool that can augment human capabilities, enabling individuals to perform a task or function better, faster, or with more precision or accuracy. Decisions, at the end of the day, are still made by humans.
For AI to have an enduring positive impact, it must be done responsibly. This means considering ethical implications, ensuring fairness and transparency, and protecting data privacy. We need to build AI systems that are not only intelligent but also trustworthy.
Maintaining an always-learning growth mindset is key and will ultimately determine whether an organization thrives or falters in the age of AI.
The writer is the director, AI Engagement and Community at Project Management Institute.
Read: