Using AI in Business: Where to Start

Daniel Bayfield - CEO, Veracity

A practical 90-day guide to using AI in business, from finding valuable use cases to choosing secure tools, setting guard rails and scaling safely.

The question business leaders are asking has changed. It was once, ‘How do we move to the cloud?’ or ‘How do we improve our cyber security?’ Today, it is increasingly, ‘How do we use AI?’.

For many organisations, the answer still feels unclear. Some employees are experimenting on their own. Others have approved tools but no shared direction. A smaller group has begun embedding AI into business processes. Wherever you sit, the best place to start is not with a long list of products, but with a real business problem.

“AI succeeds when IT enables it, and the business owns it.”

AI is already part of everyday work

AI is already helping people across sales, marketing, finance, HR, customer service and leadership. Common uses include drafting proposals and follow-up emails, generating campaign ideas, explaining reports, summarising service tickets, preparing policies, and acting as a sounding board for strategy.

These are useful entry points because they help people learn what AI can and can’t do. But the bigger shift is from asking an AI assistant for help with an isolated task to giving it the right context to support a business process or workflow.

From a general assistant to an AI that understands your business

A general AI tool knows what you tell it in a prompt. A connected business AI can, with appropriate permissions, draw on systems such as Microsoft 365, SharePoint, a CRM, finance platforms, databases, project documents and internal policies. Standards such as Model Context Protocol, or MCP, are making these connections easier to manage.

This opens the door to practical use cases such as an Internal Knowledge Assistant AI Agent. Instead of searching across several systems, an employee could ask the Agent:

•          What is our parental leave policy?
•          How do I onboard a new customer?
•          Which products are included in our premium support tier?
•          Show me the opportunities worth more than $50,000 that are expected to close this month.

The response can be grounded in approved company information and, importantly, point back to the source. Instead of employees moving between multiple systems to find answers, AI can become a gateway to the information they are already authorised to use.

Start with problems, not products

The most productive question is not, ‘What can this AI tool do?’ It is, ‘What problem are we trying to solve?’ Give AI the problem, the context and the constraints, then ask it to suggest options.

For example:

•          Our new employee onboarding process takes too long. Where could we simplify or automate it?
•          Our account managers spend too much time writing proposals. How could we reduce that effort without lowering quality?
•          Customers frequently ask us the same questions. How could we improve access to consistent answers?
•          We are struggling with staff knowledge sharing. What practical approaches should we consider?

Asked this way, AI can surface automation opportunities, workflow improvements, software you may not have considered, or ideas for a small agent. The business team remains best placed to judge which ideas are genuinely valuable.

Turning the opportunity into a practical plan

Identifying the right use case is only the beginning. How do you experiment securely, establish appropriate guard rails and move from isolated ideas to meaningful adoption?

Continue reading here: Five principles for using AI securely, a practical 90-day roadmap and the common mistakes to avoid.