At Infinity Group, we deliver complex technology projects – and complex projects always contain some level of risk.
Before any project officially starts, our sales, pre-sales and consulting teams spend weeks, sometimes months, speaking to customers. We discuss objectives, challenges, ways of working, technical requirements, resource constraints and future plans.
But this resulted in an overload of information. Every customer conversation contained clues about potential delivery risks: a passing comment about a lean internal team, concerns about training capacity, an offhand reference to wider group governance. These weren’t formal risks yet, but they were often early warning signs of challenges waiting further down the road.
Like many professional services organisations, we relied on people to capture and transfer that knowledge. But this meant they were stored in different locations, siloed away in specific documents or people’s memories.
By the time delivery teams encountered them, they were often no longer risks; they were issues. That’s what prompted us to ask a simple question: Can AI help us identify project risks before a project starts?
Creating an AI agent to discover project risk
Our AI-powered Risk Agent reviews project-related conversations, emails and documents, looking for signals that could indicate delivery risk. It then structures those findings into a standardised risk register for project teams to review. Human judgement remains central; the AI simply provides another lens through which to examine the information already available.
Before project kick-off, as early as the sales process, we can see patterns emerging.
If a customer repeatedly mentioned resource constraints, we can have an early conversation about training, testing or change management responsibilities. If discussions suggested governance dependencies or third-party approval processes, we can begin planning around them long before they became blockers.
Effectively, we created a pre-mortem process powered by the information customers were already sharing with us.
Historically, many project conversations were reactive. But now, those conversations happen much earlier.
One example is customer resourcing. It’s common for customers to underestimate the internal effort required for activities like testing, training or data migration. Traditionally, these challenges might only surface when delivery reached that stage. Today, we’re able to identify many of these risks upfront and discuss mitigation options well in advance.
Measuring the impact
The AI agent now identifies 150 project delivery risks every month. We’ve also seen an estimated 15% improvement in estimating accuracy. In professional services, where forecasting and resource planning are critical to profitability, that is a significant improvement.
On top of this, we’ve reduced the amount of unpaid remedial work caused by missed information and historic misunderstandings by around 20-30%.
Less tangible, but equally important, is the impact on project teams. Projects become stressful when teams are constantly firefighting. When risks have already been identified, discussed and assigned mitigation plans, delivery becomes calmer, more predictable and more strategic.
Lessons for professional services leaders
Having gone through this process ourselves, three lessons stand out.
-
Don’t start with AI. Start with the problem.
We weren’t looking for an excuse to deploy AI. We were trying to solve a long-standing challenge around project governance and knowledge transfer.
-
Your organisation probably already has the information it needs.
Most businesses don’t suffer from a lack of data. They struggle to connect the dots across emails, meetings, conversations and documents quickly enough to act.
-
Keep humans in the loop
The value of AI isn’t replacing experienced delivery professionals. It’s helping those professionals see patterns they might otherwise miss. Every risk we identify is still reviewed, challenged and acted upon by people.
Looking ahead
Much of the discussion around AI in professional services focuses on productivity.
That’s important, but I believe there’s another opportunity that’s often overlooked. AI can improve organisational foresight. For us, the most valuable outcome has been helping us spot problems before they happen.
Every professional services leader knows that issues become exponentially harder to solve once a project is underway. The earlier you can identify them, the better your outcomes, your customer experience and your profitability.
