Four AI agents support my day-to-day operations: here's how I got here

As I’m writing this, four AI agents keep my day-to-day operations running: a developer, an account manager, a coordinator who splits up the work, and a content agent. The whole system runs on a handful of subscriptions.

It’s worth being honest about the cost, because this is exactly where most rollouts fail. Given that I have the relevant expertise myself, for me the cost settled at a few subscription fees. So we’re not talking about one big upfront expense, but monthly services anyone can afford. So the real cost is paying for the expertise itself. In my experience, agent-based AI workflows can deliver better results in both speed and precision. Even though these tools can operate on their own, if you’re aiming for maximum quality in your business, you still set aside time to review the agents’ work. Consequently, human steps remain in the process alongside the AI solutions, or as the field calls it, “human in the loop”, but even so, we save significant time and human effort. For example, my payroll systems save roughly 900 human work-hours a year.

Where I started

It’s worth mentioning that I didn’t start from zero. Earlier, working as a sole trader in a subcontractor capacity, I was already using process-automation and related software years ago for a few recurring tasks: preparing payroll data for a staffing partner, sorting email communication, putting together reports. That was the entry level to the knowledge I have today.

As large language models (e.g. ChatGPT) advanced, I extended these processes with AI’s capabilities, which is why the range of tasks I could confidently hand off to the systems I designed and operated kept widening. What did that mean day to day? Just to name a few examples: the payroll workflow, from the moment the partner sent over the data to the point the final payroll runs were completed, was handled by an independent, automatically operating AI system. It sent a report on the tasks done, and flagged separately whenever it ran into unusual data, like a partner error. My email assistant could not only interpret and summarize my daily tasks and calendar entries, but had started proactively replying to recurring emails, always in its own name, with me strictly kept in the loop, and only after my pre-set approval. That’s the point where I felt I could really start treating AI as a colleague.

How this grew into a team

The shift in mindset happened when, instead of separate tasks, I started thinking in roles. So, colleagues: a developer, a client-relations person, a content creator, and a coordinator who ties it all together. I primarily talk to the coordinator, who then stays in touch with the rest.

This was a major step forward, because previously I had designed and built the kind of separate processes that didn’t communicate with each other. At the team level, everyone’s scope and tasks are kept separate, but they can work together, which ultimately produced more efficient work. The agents also gained something like their own personality, which helps them work with people.

The same method for other kinds of businesses and departments

The system I’ve built isn’t limited to development or content production. Generative AI itself makes it possible for this approach to be implemented in other areas too. For example, a marketing team can be built the same way: one agent writes the script, another produces the video, a third reviews it, a fourth publishes it to the company’s site. Finance or HR follow the same logic: not a single do-everything tool, but an intelligent solution that can be tailored to the client’s needs.

Conclusion

In my view, knowing the technology closely and following the trends, we can say by now that AI is capable of doing jobs people used to do, but I still say we shouldn’t see it as something that takes work away from people, but as something that complements it. In upcoming posts I’ll also talk about the pitfalls of artificial intelligence and how to avoid them. And about how a business can meet its legal obligations, such as the EU’s AI Act.

If you’d like to see which role in your business would be the first worth handing to an AI agent, request the free assessment.