Published on November 18, 2026
How ainavio uses AI agents in its own software development: with ATM v1, agents handle research and implementation, while people decide, review, and accept. The goal is faster answers to customer requests without compromising on quality.

A customer request often starts with a simple question: "Can the product do this?"
Answering it well takes more than writing code. We need to understand the customer's situation, decide what will actually solve the problem, make the change, test it, and check that the result meets the original need. Every step matters, and every step can become a bottleneck when the team is busy.
At ainavio, we use AI to move through this process more effectively. The goal is to give the team more capacity to listen, respond, and deliver useful improvements, while people stay responsible for decisions and product quality.
Over the past few years, AI coding tools have become capable assistants. But most of them still depend on a person sitting at a computer, giving instructions and waiting for the result. When that person moves on to another task, the work stops.
We wanted to try a different approach: what if AI agents took part in the team's existing workflow, rather than waiting inside one developer's workspace?
That idea became ATM v1, our Agentic Task Management system. It connects AI agents to the way our team already plans, develops, and reviews work. Agents pick up suitable tasks, carry out their part, and bring the results back for review. They keep working while our engineers focus on other priorities.
For the team, this means more work can move forward in parallel. For customers, the intended benefit is a more responsive product team, with less time lost waiting between steps.
Customer feedback is only valuable if we understand the problem behind it. A request might describe a desired feature, a confusing workflow, or an issue that needs investigation: an additional attribute a marketplace suddenly requires, for example, or an import that behaves differently than expected. Before implementation begins, we need enough context to decide what should happen next.
ATM helps the team clarify a request, research relevant information, and prepare it for review. Our colleagues stay involved: the person who submitted the request confirms the details, and the product owner decides how to handle and prioritize it.
From there, the work stays connected to the original request. The team can follow its progress and return to the customer's actual need when evaluating the result.
Speed matters, but a fast change is only useful if customers can rely on it. That is why we designed ATM's workflow with checks and human decisions throughout.
We treat an agent's statement that a task is finished as a claim to be checked. Implementation is reviewed separately, and testing produces evidence that people can assess. People approve plans, review changes, and accept results. When an agent runs into a question it cannot resolve, it passes the context back to the team and continues once a decision has been made.
This separation is important. It lets AI contribute to more of the work without making the same agent the sole judge of its own output. And it gives our engineers and product team the opportunity to catch mistakes, clarify requirements, and improve the result.
AI can generate a lot of activity. Pull requests and completed tasks are useful indicators of capacity, but they do not tell the whole story.
We also look at how long a customer request takes to become a validated improvement, how much rework is needed, and whether the change solves the problem it was meant to address. Quality and customer outcomes belong alongside speed in any conversation about engineering performance.
We keep developing this approach as we learn. ATM v1 is a practical step toward a team in which people and AI agents each contribute where they are most useful: agents help move work forward, while people provide judgment, direction, and accountability.
For our customers, the ambition is straightforward: a product that listens, learns, and evolves with their needs.
Talk to us about your range, your channels and your legacy systems. We will show you where PIM with AI fits into your setup.
Conclusion: AI makes ainavio's product team faster only where it is embedded in a workflow with clear responsibilities. Agents take on tasks and keep work moving; people decide, review, and accept. Success is measured not by the number of pull requests, but by how quickly a customer request becomes a validated improvement.