Dive Brief:
- Most enterprises are exceeding their AI budgets as agents and agentic software development usage boomed over the last year, according to a McKinsey report published last week. The research company collected data from search engine queries, news articles, patents, research publications, equity investment and talent demand for its report.
- While agentic software development is becoming standard practice in enterprises, its ubiquity isn’t yet showing results. Only one-quarter of companies said the tools have achieved meaningful acceleration in their product development life cycles, and in 30% of companies, productivity fell after teams began using agentic AI, the report found.
- “To capture the value from these tools, the software development life cycle operating model has to shift, including moving toward smaller, highly leveraged teams that supervise agents through execution,” Martin Harrysson, a senior partner at McKinsey, said in the report.
Dive Insight:
Agentic AI systems and agents are changing the fabric of enterprise workflows — humans are working alongside agents, retooling operating models companies have long relied on.
Investment in agentic software development is set to grow more than twelve times from 2025 to 2026, McKinsey found. Global AI spend overall is forecasted to total $2.7 trillion this year, representing a 49.5% increase year over year, according to recent Gartner data.
The adoption of coding agents is transforming engineering roles, as the technology shifts job responsibilities to managing AI outputs. Some organizations are beginning to deploy after-hours workflows, in which agents work asynchronously overnight or across weekends, McKinsey found.
Developer acceptance of agentic software tools is a constraint on the tool’s success, according to McKinsey — nearly half of developers worldwide said they actively distrust AI tools’ accuracy.
Yet enterprises continue to pursue adoption with productivity gains in mind. Work that used to happen through two-week sprints can now be done in a continuous loop of drafting, testing, debugging and documentation between agents and human reviewers, Prakhar Dixit, a partner at McKinsey, said in the report.
“The leadership challenge is to increase speed without weakening maintainability, quality, or control,” Dixit said.
AI providers are competing for enterprise spending as they look to deploy agentic coding tools. SpaceXAI purchased AI coding agent Cursor’s maker Anysphere in August for $60 billion, a flow investment into the AI software development market. Anthropic’s Claude Code and OpenAI’s Codex, which both launched more than a year ago, are also major players in the space.
Autonomous-agent deployment is creating demand for structured knowledge graphs, advanced repository indexing and enterprise context systems that can help agents navigate code, policy, product history and operational rules. Enterprises with clear documentation may have an easier time finding value from their AI investments, McKinsey said.
Enterprises will likely continue to develop the relationships between their human workers and the agentic tools they deploy. Some are beginning to treat agents and software development economics as a mix of labor cost and AI-usage cost, the report said.
“The defining question of the agentic era is not how autonomous agents can become but how much autonomy the enterprise can safely absorb,” Oana Cheta, a partner at McKinsey, said in the report.
