Sam Altman just laid out his vision for where AI is headed next, and it goes well beyond the chatbots we’ve all been arguing with for the past two years. The OpenAI CEO describes a “third wave” of artificial intelligence built around three pillars: chatbots, coding agents, and persistent AI agents that will essentially function as staff, co-workers, and colleagues.

From tools to teammates

Altman isn’t talking about AI as a productivity tool you open in a browser tab. He’s describing systems that operate autonomously, handling complex tasks independently, with persistence, meaning they don’t forget what they were doing when you close your laptop.

The first wave of AI was predictive. Think recommendation algorithms and spam filters. The second wave was generative, the ChatGPT era of creating text, images, and code on demand. This proposed third wave is agentic: AI that doesn’t just respond to prompts but takes initiative, manages workflows, and executes multi-step processes without constant human oversight.

Altman has predicted that the first AI agents will join the workforce in 2025, meaningfully boosting company output. Salesforce CEO Marc Benioff has set a target of deploying one billion Agentforce agents by the end of 2025.

What this means for crypto and tech infrastructure

OpenAI partnered with Paradigm to release EVMbench in February 2026, a benchmark designed to assess AI’s capability in securing crypto smart contracts. That’s a direct play at one of crypto’s most persistent vulnerabilities: the billions of dollars lost annually to smart contract exploits and bugs.

The crypto coverage around OpenAI’s developments has remained focused on practical applications like security rather than speculative token launches.

The productivity thesis and its market ripple effects

For investors, the real signal in Altman’s comments isn’t about any single product announcement. It’s about a fundamental shift in how companies think about headcount, productivity, and operational leverage.

For the crypto market specifically, persistent AI agents that can autonomously manage wallets, execute trades, or interact with DeFi protocols would need robust identity, authorization, and audit mechanisms. Blockchain-based solutions for agent identity and accountability could become a meaningful use case as genuine infrastructure.

Traders watching the AI narrative should pay attention to enterprise adoption metrics rather than headline announcements. When companies start reporting AI agents as a line item in productivity gains during earnings calls, that’s the signal that the third wave has actually arrived.

The intersection of persistent AI agents and crypto security tooling, like EVMbench, suggests that the most durable investment thesis here isn’t about picking winners in the agent race itself. It’s about identifying the picks-and-shovels plays: the infrastructure, security, and compliance layers that every agent-powered enterprise will eventually need, regardless of which AI provider wins the modeling wars.