By 2029, most production code may be written by AI, and the engineers who remain will be judged on whether that code is right. Alexey Tulia, Executive Leader at Coinspaid Dev, shared that forecast at Tech Race Summit 2026 and explained what companies should do about it now.
As covered by Hackread, Tulia took part in the AI Impact in Engineering panel, where the conversation turned to AI systems that no longer just suggest but act, working directly through company infrastructure. His remarks moved from long-term predictions to the decisions CTOs face this year.
The 2029 picture
Tulia’s outlook is straightforward. Engineering teams will get smaller, and each will be responsible for a larger slice of the business. AI will generate the bulk of the code that ships. As the cost of producing software falls, companies will also bring in more vendors and more AI-built systems, which adds complexity rather than removing it.
That combination raises the value of one skill above the rest. “I think technical judgment becomes even more important,” Tulia said. The CTO of the future, in his view, still needs deep technical expertise. What grows alongside it is business understanding and the ability to verify work that nobody on the team typed by hand.
Engineers move closer to outcomes
The shift is already visible at the level of individual engineers. With AI speeding up coding and prototyping, Tulia argued, engineers gain time to understand the underlying business problem and to stay involved until the result is running in production.
Leaders play a part here. Teams need to know the business context of their work and the outcome that counts as success. Counting lines of code makes little sense in this setting. Tulia suggested looking at correctness, maintainability, security and operational performance to gauge whether a team is productive.
The accountability question
Everything above leads to a harder issue. Companies are used to AI that drafts and analyses. The next phase gives agents access to live systems, from sensitive data to deployment pipelines, so they can make changes themselves. The timing matters: across the AI industry, researchers and executives are openly questioning whether safety work can keep up with how quickly capabilities are advancing. Tulia focused on what that tension means inside an ordinary engineering organisation.
“The more authority we give machines, the more important accountability becomes,” he said.
Picture an agent that has prepared a change and has the permissions to push it to production. Can it go ahead without a person approving it? If the release takes something down, who is accountable? Tulia’s answer was that such autonomy should only come after the company has built permission controls and audit logs, and has the means to halt the agent and recover from a bad deployment. Authority has to be clearly defined, and a human has to stay responsible for the result.
Making room to be wrong
How should a CTO prepare for a future that is hard to forecast? Tulia’s approach is to keep the cost of mistakes low. “I don’t need to predict the future perfectly. I need to make being wrong cheap,” he said.
In practice, that means linking AI spending to a real organisational need and putting money into the groundwork that makes change safe:
- strong, well-maintained APIs;
- reliable, accessible data;
- automated testing and observability;
- security built into the stack;
- flexible architecture with limited vendor lock-in.
Several of these produce little revenue in the short term. Their payoff arrives when a provider must be swapped out or a system rethought because an assumption proved wrong. Tulia also stressed spare capacity. A roadmap that uses up every resource leaves no space to test a promising tool or respond when priorities shift.
For engineering leaders, the conclusion is concrete. Ownership and safeguards should be settled before AI agents are allowed anywhere near critical production systems, not after the first incident.
About Coinspaid Dev
Coinspaid Dev is an independently owned and operated software engineering company specialising in blockchain infrastructure development. Its more than 120 engineers bring over 11 years of industry experience across software engineering, infrastructure, security and R&D, building distributed systems that operate on more than 20 blockchain networks.
