Agentic AI and the Code Carnival That Forgot Its Ticket

The rapid adoption of agentic AI tools sounds like the moment a factory installs a robotic barista: everyone expects faster lattes, prettier foam, and maybe a small existential crisis. But the latest McKinsey Technology Trends Outlook 2026 suggests the steam may be escaping in the wrong direction. Productivity actually declined in nearly thirty percent of companies after teams started using autonomous AI systems.
The report says trouble appears when firms drop the new tools on existing workflows like fancy paint on a cracked wall. Agentic AI can automate pieces of software development, but the resulting gains have varied wildly. Without a systematic approach, agentic software development can produce unintended outcomes. Almost everyone, it seems, is vibe coding, yet that does not always lead to value.
I once watched a promising team install an agentic copilot like a jetpack on a bicycle. The bike went faster, sure, but no one changed the route, the brakes, or the fact that they were still carrying a refrigerator. The result was code galore and release dread. That anecdote feels familiar. I have personally stared at a chatbot suggested function that looked brilliant until it politely invented a method for a library that had never met it. The thrill was real. The debugging was spiritual.
More code does not mean more products, McKinsey notes. A rise in coding activity can fail to become useful output or faster business growth. In one study, AI tools increased coding activity by 180 percent, but shipped releases rose by only 30 percent. That is the software equivalent of filling a shopping cart with twenty pizzas and paying for one slice.
There is also a trust problem wearing a developer hoodie. Around 46 percent of developers globally said they actively distrust the accuracy of AI tools, compared with 33 percent who said they trust them. Only 3 percent said they highly trust AI-generated outputs. Developer acceptance remains a meaningful constraint, and for companies hoping to harvest agentic gains, that gap matters like a missing step on a staircase people are expected to sprint down.
Yet investment surges anyway, because optimism can be the most caffeinated employee in the building. Equity investment in companies developing agentic software rose from negligible levels through 2023 to around 5 billion dollars in 2025, then crossed 61 billion dollars in the first half of 2026, largely because of a 60 billion dollar acquisition of Cursor. Job postings related to the technology rose 221 percent between 2024 and 2025.
The wider tech sector is also spending at a pace that makes a teenager on a sale website look prudent. Investment in AI infrastructure doubled in a year, while energy technologies attracted nearly 200 billion dollars in 2025 as companies chased the power AI needs. This breakneck pace creates challenges for organisations racing to deploy without proven road maps. The fix is less about buying more robots and more about teaching the humans when to press start. The future is fast, but not necessarily better or faster.