AI is evolving fast, and so are the ways we use it. While tools like Copilot have already transformed how we create, communicate and collaborate, the next wave is here: AI agents.
Unlike traditional AI tools that assist with isolated tasks, AI agents are designed to take on more complex, multi-step processes. They’re proactive, context-aware and capable of interacting with systems and data to help us deliver smarter and faster.
At Infinity Group, we’ve started exploring how these agents can support our teams. Our goal is simple: reduce risk, remove ambiguity and give our consultants more time to focus on the work that truly requires human insight.
In this blog, we’ll share how we’re using AI agents internally within our project operations – and how you can get inspired to build your own.
Why we’re using AI agents
AI agents represent a significant shift in how organisations can approach delivery, governance and collaboration, moving from conversational AI to AI that can actually execute tasks.
As interest in agents grows, many businesses are asking not just if they should use them, but how. As a partner, it’s crucial for us to lead the charge when it comes to innovation. This then enables us to best give practical advice to our clients and inspire them with high-value use cases. So, it seemed only fitting that we implement AI agents internally – including within our project delivery function.
Like many delivery teams, we’ve faced recurring challenges: ambiguous requirements, misaligned expectations and time-consuming manual tasks that slow down progress and increase risk.
AI agents offered a practical solution. By automating the parts of the process that are prone to error, we saw an opportunity to improve project outcomes without compromising the human expertise that drives them.
The motivation was twofold:
- To reduce delivery risk by identifying gaps and inconsistencies early, before they become costly
- To support our consultants by removing repetitive tasks and giving them a stronger starting point for design and decision-making
How we found our AI agent use cases
One of the most common questions we hear is: “How do you know where to apply AI agents?” It starts with understanding the work itself.
Our team began by looking closely at the delivery process, identifying areas that are repetitive, prone to ambiguity, or difficult to scale. These are often the tasks that slow projects down or create friction between teams and clients. From there, we examined whether an AI agent could help.
To support this thinking, we also used AI itself to generate ideas. By prompting tools like Copilot with questions such as “What could AI support in an implementation process?” (broken down by role and task), we received a range of suggestions. Some were already in place, like using Teams to capture meeting notes. Others sparked new ideas and helped us spot gaps we hadn’t considered.
The key is to break down roles and workflows into every step. Much of the knowledge consultants carry is tacit; held in their heads or passed on informally. But if a human can access that information, so can an AI agent. By mapping where that knowledge comes from (often libraries, documentation, or past projects), we can design agents that support those same decisions and processes.
It’s crucial to note that this approach isn’t about replacing expertise. Instead, the focus is surfacing it, scaling it and making it more consistent across teams.
The AI agent use cases we’re testing
Knowing the areas where AI agents could add value and reclaim time for our project teams, we’ve created a series of custom agents tailored to our workflows and project needs.
Here are some of the agents we’ve created so far:
Pre-mortem agent
This agent is designed to analyse pre-sales documentation and internal recordings. It generates a comprehensive risk profile for each project before it begins, highlighting potential issues and enabling both our team and the client to prepare effectively. By doing so, it forms the foundation for our project risk register, giving us a valuable head start on governance and planning.
Workflow support agent
Project delivery often involves multiple moving parts – internal handovers, documentation updates, stakeholder communications and more. The workflow support agent was created to help teams stay aligned and efficient throughout these processes.
The agent acts as a behind-the-scenes assistant, monitoring internal workflows and prompting action where needed. Whether it’s ensuring key documents are in place, nudging teams to complete handovers or flagging missing inputs, it helps maintain momentum and consistency across delivery stages.
Design and requirements agents
Requirements gathering and solution design are two of the most critical and error-prone phases of any project. Misunderstandings at this stage can lead to costly rework and misaligned expectations later.
To address this, we’ve developed agents that support both business analysts and solution architects. These agents help:
- Validate requirements for clarity and completeness
- Cross-check inputs across departments to ensure alignment
- Compare requirements against Microsoft’s standard processes
- Suggest design approaches based on best practices
For example, if a requirement says “run a report,” the agent prompts for specifics: What kind of report? When should it run? What should it include? This helps eliminate ambiguity and ensures everyone is working from the same understanding.
Together, these agents act as a quality layer, helping teams catch issues early, align across functions and build designs that are both robust and relevant. Most crucially, it saves them time for the work that truly makes a difference to clients and utilises their skills.
Next steps
Our work with AI agents is just beginning. As we continue to build and refine these tools, our focus remains on practical impact, supporting delivery teams, improving project outcomes and helping our people do their best work. Human expertise remains central to everything we do, but agents are tools to help us work smarter.
We’re continuing to develop new agents to tackle challenges and add value across workflows – in the project team and beyond. As agents are deployed, we’ll track how they affect delivery quality, efficiency and team experience. This will help us refine their design and ensure they’re solving the right problems.
As platforms like Copilot evolve, we expect natural language agent creation to become a reality. That shift will open new possibilities for how we design, deploy, and scale AI support across the business.
And most crucially, we’ll continue to share insights and ideas that inspire our clients to make the most of agentic AI in their organisations.
Ready to explore AI agents for yourself?
AI agents are already helping us deliver projects with greater clarity, consistency and confidence. By focusing on practical use cases and people-first design, we’re building tools that support teams. And as the technology evolves, so will the opportunities to innovate.
If you’re curious about how AI agents could support your own delivery model, we’ve put together a guide to help you explore what’s possible. Find out everything you need to know about Copilot Agents, alongside example use cases and best practices.
