Practical thinking on
AI and automation
for businesses like yours.

AI agents most commonly fail not because the technology is broken, but because the business did not define the edge cases the agent would encounter, did not surface the institutional knowledge the process depends on, did not build a framework for knowing when the agent should involve a human rather than continue acting, and had no way of measuring whether the agent was performing correctly over time. Each of those gaps is preventable, and all of them are easier to address before deployment than after.

The most important thing to look for in a Power Automate consultant is whether they ask about your business before they propose a solution, because technical competence with the platform is widely available and tells you relatively little about whether they will solve the right problem for your business.

Power Automate consultants help small businesses get real results from a workflow tool most are already paying for through their existing Microsoft 365 subscriptions. The basic licence is included in most M365 plans, the premium licences unlock significantly more, and most businesses never have anyone work out what either could actually do for them. Whether you get real value depends almost entirely on whether the consultant starts with your business problems or with the platform.

Small businesses can use AI agents for business operations by starting with the agent capabilities already built into the tools they use, identifying tasks where language and judgment are involved and errors are recoverable, and establishing how they will measure performance before deploying anything. The case for agents is real, but most of the public debate about them is happening at a level of complexity that is irrelevant to a small business, and understanding that distinction is what lets you make sensible decisions rather than expensive ones.

Most small businesses need four categories of AI tools: tools that help individuals work faster, workflow automation that replaces manual processes, AI agents that handle multi-step tasks with reduced human involvement, and custom-built AI for processes the off-the-shelf options do not cover. Most businesses only use the first category, which is why AI investment so often produces subscriptions without producing change.

Start with the process that is costing your business the most in lost time, failed handovers, or client impact. Before choosing any tool, map your most operationally critical workflows and test each against three questions: how often does it happen, what breaks when it goes wrong, and is the process stable enough to automate. That narrows the field to genuine priorities.