At a Glance

  • Half of software engineering teams see agentic AI as their top investment priority this year, and 84% say it will become the dominant investment within three years, according to a report released Tuesday by SoftServe. The report, conducted by MIT Technology Review, surveyed 300 CIOs, CTOs, and other technology leaders from December through January.
  • Despite teams' efforts to deploy the technology, integration of agents with existing systems and the cost of compute resources remain major barriers to AI adoption, the survey found. Only 12% of teams report widespread use of agentic AI in their organizations, according to the report.
  • AI experts in the report say that once fully integrated, agentic AI will become prevalent in areas such as code generation, testing, refactoring, and project deployment. Within about three years, early adopter areas like software are expected to see agentic engineering involved in overall lifecycle management, the report says.

Deep Dive

Software development teams are early adopters of agentic AI, but as tools mature, their application has recently expanded across multiple industries. According to the SoftServe report, most organizations expect to increase investment in workflows over the next few years.

However, the cost of integrating these tools is a significant consideration for technology leaders. The State of Data Analytics Engineering Report released today by Dbt Labs, which surveyed 363 data practitioners and leaders from December through January, highlights cost as a particularly prominent issue.

More than half of respondents said they are increasing warehouse and compute spending to meet agentic demands, while only 36% of leaders have increased team budgets. The Dbt report found that while technical integration has become easier for many teams, trust in data quality remains a concern. Dan Poppy, senior content manager at Dbt Labs, said in an email that clear attribution, validated outputs, and documented data models are foundational for data teams.

"Data leaders who treat governance as infrastructure will be able to say 'yes' to AI faster and more reliably than others," Poppy said.

Technology leaders expect productivity gains from agentic AI to take time, with most respondents in the MIT and SoftServe report expecting slight or moderate improvements within the next two years. Leaders foresee AI agents accelerating teams' delivery of software projects and hope they will manage product development and software development lifecycles over time.

But Poppy said AI failures will present challenges different from those technology teams have faced before. "When traditional systems crash, you can tell. If an executive asks AI about new annual recurring revenue (ARR) over the past 30 days and the AI system fails, it could produce a plausible and confident wrong answer that reaches a board presentation or a customer before anyone notices," he said.