Smaller Teams, More AI Code: Alexey Tulia Explains What Engineering Leaders Must Prepare For

By 2029, AI could be writing the majority of production code while leaner engineering teams look after larger parts of the business. Alexey Tulia, Executive Leader at Coinspaid Dev, says this future will depend less on the tools themselves and more on how clearly companies assign authority and responsibility.

As covered by DEV Community, Tulia outlined these expectations at Tech Race Summit 2026 in Warsaw, speaking on the AI Impact in Engineering panel. The discussion centered on how the jobs of engineers and CTOs are evolving as AI moves beyond helping with drafts and analysis toward systems that can act directly through company infrastructure.

For individual engineers, the change is already visible. AI tools make writing code and building prototypes much faster, and Tulia believes the time saved should go into understanding the business problem a feature is meant to solve and following that feature all the way into production. That kind of ownership needs support from above. Leaders are expected to explain the business context to their teams and agree on the result each project should deliver. Once that is in place, output can be judged by how correct, maintainable, secure and stable the software is in operation. Counting lines of code tells a manager very little in a world where a model can generate thousands of them in minutes.

The harder issue appears when AI stops advising and starts doing. The next wave of adoption will link agents to live environments, giving them reach into sensitive data and deployment pipelines. Tulia summed up the tension in one line: “The more authority we give machines, the more important accountability becomes.” Imagine an agent that can write a change and push it to production by itself. A company has to decide in advance whether that push needs a human to approve it and who is accountable if the deployment goes wrong. Tulia’s answer is that such access should come only after the organization has built the controls around it. Permissions must limit what the agent can reach, audit logs must capture what it did, and teams need both a way to stop the agent and a tested path to recover from a broken release. Autonomy in production, as he framed it, is only safe when authority is defined and a named person remains responsible.

This is also why Tulia pushes CTOs to think carefully about where AI budgets go. Each investment, he said, should answer a concrete need inside the organization. His list of priorities is practical: strong APIs, trustworthy data, automated testing, observability, security and an architecture flexible enough to absorb change. None of these produce dramatic headlines, and efforts to reduce dependence on a single vendor seldom add revenue in the short term. Their value shows up later, when a provider has to be replaced or a system rebuilt because the original assumptions no longer fit. Teams also need slack in their plans, since a roadmap with every hour already allocated leaves nothing for testing a promising new tool or adjusting to a sudden change in priorities. “I don’t need to predict the future perfectly. I need to make being wrong cheap,” Tulia said.

As building software becomes easier, companies will bring in more vendors and more AI-generated systems, and someone will have to evaluate all of them. That is where Tulia sees the CTO role heading: toward a leader who combines serious technical depth with a clear understanding of the business. “I think technical judgment becomes even more important,” he said. His immediate advice for engineering leaders is to agree on safeguards and ownership now, before any AI agent is handed the keys to critical production systems.

Tulia leads at Coinspaid Dev, an independently owned and operated software engineering company focused on blockchain infrastructure development. With a team of more than 120 engineers and over 11 years in the industry, the company combines software engineering, infrastructure, security and R&D expertise, and its teams have built distributed systems and blockchain infrastructure running on more than 20 blockchain networks.