Leading firms are deploying personalized AI agents at a massive scale. McKinsey already has 25,000 agents for its 40,000 employees and expects to reach parity within the year. The key skill is shifting from doing work to conducting an 'orchestra' of agents.

The theoretical discussion about AI and job loss is becoming reality. One startup founder plans to replace 70% of his team (50 people) with "agent swarms"—interconnected AI agents that handle specific functions managed by a master agent. This indicates job displacement may be more rapid and widespread than anticipated.

McKinsey's global managing partner now considers AI agents part of the company's headcount. The firm rapidly scaled from 3,000 to 20,000 agents in just 18 months, viewing them as essential 'employees' that augment their human workforce, and expects to reach a 1:1 human-to-agent ratio by 2026.

Contrary to the view that useful AI agents are a decade away, Andrew Ng asserts that agentic workflows are already solving complex business problems. He cites examples from his portfolio in tariff compliance and legal document processing that would be impossible without current agentic AI systems.

Union Square Ventures' Nick Grossman argues the dominant mental model for AI is shifting. Instead of thinking of agents as personified 'employees' to hire, we should see them as cloud infrastructure. Businesses will programmatically spawn and orchestrate thousands of agents as integral parts of their core systems.

The number of AI agents will soon vastly exceed human employees. This requires a fundamental shift in software development, prioritizing API-first design, reliability, and machine-to-machine interaction over traditional human-centric user interfaces.

Drawing a parallel to the microservices boom, enterprises will soon deploy thousands of AI agents, creating immense operational complexity. The most valuable future products will be those that, like Datadog for microservices, provide governance, monitoring, and orchestration for this sprawling agentic workforce.

The tangible utility of agentic tools like Claude Code has reversed the "AI bubble" fear for many experts. They now believe we are "underbuilt" for the necessary compute. This shift is because agents, unlike simple chatbots, are designed for continuous, long-term tasks, creating a massive, sustained demand for inference that current infrastructure can't support.

The next wave of AI adoption involves 'agentic' workflows, where AI performs complex tasks autonomously. This shift from simple queries to agentic use is expected to increase token consumption by approximately 10x per task. This will drive a massive explosion in compute demand across all knowledge-work industries, not just coding.