The bigger takeaway from this study is that product managers and domain experts, people who can direct and guide what the machine does, are going to be in far greater demand going forward | Photo Credit: Getty Images

Here’s some good news for those who are overtly worried about the threat of being replaced by AI. A recent survey conducted by the AI company Anthropic has found that domain expertise has a clear edge. The more understanding a worker brings to an agent, the higher-quality work the agent can do.

Agentic coding has arrived — from Microsoft to TCS, and even small start-ups have begun deploying Agents to handle a wide range of tasks. That the share of GitHub projects with coding agent activity has more than doubled since late 2025 reflects the sharp spurt in model-assisted coding. 

But here’s a catch. Knowing how to code doesn’t automatically give one an edge in conducting successful coding development sessions on Claude Code, resulting in the development of agentic tools. It is the domain expertise of those who are interacting with the model is the differentiator, according to the latest Anthropic study.

It studied over four lakh Claude Code sessions conducted by 2.35 lakh people between October 2025 and April 2026, in which people make most of the planning decisions (what to do), and Claude makes most of the execution decisions (how to do it).

“The greater domain expertise a person brings to a session, the more work Claude does per instruction. The more domain expertise a person has, the more often the session ends in success,” the AI company said in its latest study.

“Claude is handling more complex and more valuable tasks. At the same time, there remains a clear division of labour in agentic coding: People decide what to build, and the agent decides how to build it,” it said.

Domain expertise, and not coding proficiency, amplifies effective use of the tool. Domain experts succeed more often and recover more easily from errors and misunderstandings, while those with little experience are more likely to give up when they struggle to achieve the expected outcome.

Paramdeep Singh, Co-Founder of Shorthills AI, likened the human-AI collaboration to directing a movie. 

“The human plays the director, giving instructions, while AI handles the actual execution, the acting. Anthropic’s study on AI-assisted coding reflects exactly this dynamic. The data show that work is best done when guided by an expert human,” he said.

In practical terms, the human defines the brief and the AI writes the code. Routine, tedious tasks get automated — Claude or any other AI handles the programming.

“But when it comes to the product itself, it is the product manager, the person with domain expertise, who plays the director’s role and tells the AI exactly what needs to be done,” he added.

Citing the example of the task of closing the books. “That is an accounting process where an AI agent can help automate reconciliation. But the domain expert, whether that is a product manager or an accountant, is the one who tells the AI what guardrails to put in place,” he said.

The bigger takeaway from this study is that product managers and domain experts, people who can direct and guide what the machine does, are going to be in far greater demand going forward.

Published on June 17, 2026