As noted by The Tech Portal, Tulia shared these views at Tech Race Summit 2026 as part of the AI Impact in Engineering panel. He focused on a transition many companies are only beginning to face: AI moving from a helper that drafts and analyses to a system with direct access to infrastructure, sensitive data and deployment pipelines.
A large part of his talk dealt with where technology budgets should go. Tulia advised CTOs to link every AI investment to a concrete need inside the organisation. The areas he put first were solid APIs, dependable data, automated testing, observability, security and an architecture that can be adjusted without major rework. In his view, these are what allow a company to bring in new technology without putting its operations at risk. Efforts to limit vendor lock-in belong in the same category. They seldom produce revenue right away, but they make it far easier to switch providers or rethink a system once earlier assumptions stop holding. “I don’t need to predict the future perfectly. I need to make being wrong cheap,” he said.
He added that a roadmap filled to the last hour leaves engineering teams unable to try out a promising tool or respond when business priorities move. Keeping some capacity free for experiments is, for him, part of the same strategy.
Tulia also addressed how the work of individual engineers is evolving. With AI accelerating coding and prototyping, he said, engineers gain time to understand the business problem they are solving and to stay with their work after it reaches production. Managers can encourage this by explaining the business context and stating what outcome they expect. Output volume becomes a poor yardstick in that setting, so he suggested judging teams on the correctness, maintainability, security and operational performance of what they deliver.
The question of accountability runs through all of this. Tulia described an agent capable of preparing a change and releasing it to production, and asked whether it should ever do so without a person signing off, and who would carry the responsibility if the release failed. Before any such access is granted, he said, an organisation needs controls over permissions, logs that record what the agent did, a way to halt it and a tested process for recovering from a bad deployment. Wider autonomy in production, he argued, calls for sharper definitions of authority and human responsibility. “The more authority we give machines, the more important accountability becomes,” Tulia said.
Looking to 2029, he predicted that leaner engineering teams will cover broader scopes of work and that AI will write most production code, which will raise the stakes for verification and technical judgment. The CTO will still need strong technical depth together with a grasp of the business, particularly as simpler development tools bring more vendors and AI-built systems into companies. “I think technical judgment becomes even more important,” he said.
Coinspaid Dev is an independent software engineering company focused on blockchain infrastructure. It employs more than 120 engineers, has over 11 years of experience in the industry, and its teams build distributed systems running on more than 20 blockchain networks. For leaders in this space and beyond, Tulia’s message was that safeguards and clear ownership need to be in place before AI agents are allowed near critical production systems.
