Every week seems to bring a new AI model, assistant, feature ad nauseam and lots of companies have spent the past couple of years asking: "How do we use AI?"

‍While that's a reasonable question, it's probably not that useful.

You’ll see a lot more value from taking a step back and rather than looking at the technology first, look at what you do and what you could do better.

A more useful question to ask is this:

Which business processes consume time, follow clear rules and happen repeatedly?

‍We’re all familiar with AI assistants now: ask a question, upload a document or draft some text and get something back. While this is undoubtedly useful, we're moving beyond assistance and into execution.

Instead of helping you complete a task, AI can increasingly carry out parts of the task itself e.g. gathering information, updating records, triggering actions, passing work to the next person in the process.

Many of the most valuable uses of AI help with improving business processes.

That's why technology like Model Context Protocol (MCP) matters.
‍
Despite the rather uninspiring name, MCP is quickly becoming one of the most important developments in practical AI. You could describe it as "USB-C for AI" because it provides a standard way for AI tools to connect to data, systems and actions rather than relying on countless one-off, bespoke integrations.

Let’s look at a couple of examples:

Consider a year-end accounts preparation process. Information may sit across your practice management system, document management platform, email conversations and working papers. An AI agent could gather the relevant information, identify missing documents, update internal tracking records and route exceptions to the right person for review. The process still has an owner and so human oversight remains. It just removes the manual effort.

Or take an AML or client onboarding process. Rather than manually checking multiple systems, chasing information and updating records, AI can retrieve information from the relevant sources, flag missing requirements, trigger follow-up actions and keep the process moving.

Many firms are still treating AI as a productivity tool for individual users.

A bigger impact will come from looking at the operational friction that exists across your firm and identifying where information gets stuck, duplicated or delayed.

And most importantly, this doesn't require abandoning governance. In fact, some of the most promising developments are built around preserving it.

AI can work within existing permissions. Audit trails remain intact. Approval processes can still require human sign-off and rather than bypassing controls, the technology can operate within them.

We think it’s most sensible to stop chasing the flashiest new stuff and shoehorning a way to use it and identify where AI can genuinely remove effort without increasing risk.

So instead of asking:

"What can AI do?"
‍
Try asking:

"What work are our people still doing manually that they shouldn't have to?"
‍
As both a Microsoft partner and an iManage partner, Quorum is helping organisations explore how technologies like Microsoft 365, iManage, and beyond can work with AI together to create practical, governed business processes.

If you're exploring MCP, Copilot Studio, AI agents or process automation, start with one process. Map the systems involved. Identify the bottlenecks. Then ask what could be delegated safely and effectively.

You might find that your biggest opportunity comes from fixing the processes that have been frustrating your people for years!

If you'd like to explore where AI, MCP and connected business systems could deliver measurable value in your firm, we’re offering a free workshop or discovery session - https://bit.ly/AIConsultationRequest