The idea of a single general AI able to do anything is appealing, but the path proving more concrete and effective in small businesses is a different one: specialised agents, each designed for a particular task, built into the specific software that task already lives in.
Specialisation beats generality
An agent specialised in quoting knows the calculation formulas thoroughly, the structure of the company's data, the specific language of the trade. In practice that depth is more useful than a general assistant able to talk about anything and expert in nothing.
Built into the work you already do
The real value lies not in having an AI agent as a separate tool to consult on the side, but in having it inside the software you already use every day — where it can read and act on real data, rather than discuss it in the abstract in an isolated chat window.
What changes for people in the office
The most concrete change is not tasks disappearing but time shifting: fewer hours spent on repetitive sums and hunting for information, more time on negotiation, customer relationships and decisions that genuinely need human judgement — the one part of the job no agent, however capable, can take over.