[Digital Today reporter Jinju Hong] Cryptocurrency exchange Coinbase has begun piloting artificial intelligence (AI) agents to support internal work, accelerating an organisation-wide shift to an "AI-native" model.

Cointelegraph, a blockchain media outlet, reported on April 20 that Coinbase CEO Brian Armstrong (브라이언 암스트롱) said the company is introducing AI agents into its Slack and email systems to integrate AI deeply into work processes.

Armstrong disclosed that two AI agents have already been deployed. The agents were designed based on former company executives, and the company plans to make it easy for any employee to create a customised agent for themselves or their team. He predicted that "at some point, there will be more agents than human employees."

One of the agents disclosed is "Fred," based on co-founder Fred Ehrsam. It is an executive-style agent for strategy, designed to help set internal priorities and organise strategic direction. The company explained that it aims to provide executive-level feedback.

Another agent is "Balaji," modelled on former chief technology officer Balaji Srinivasan. Coinbase introduced Balaji as an agent of chaos and creativity. It aims to spur innovation by assigning it to challenge existing assumptions and support thinking outside the box.

Coinbase's strategy to expand AI has already surfaced several times. Armstrong set a goal last September for AI to write more than half of the company's code, and has cited converting an organisation of about 4,000 people to an "AI-native" environment as a key task.

The company is also preparing payment infrastructure based on AI agents. In May 2025, it unveiled the "x402" protocol supporting both cryptocurrency and fiat currency payments, moving to build an environment in which AI can carry out transactions and payments.

Similar forecasts are spreading across the industry. Circle CEO Jeremy Allaire said billions of AI agents will trade on-chain within 3 to 5 years, and former Binance CEO Changpeng Zhao (CZ) has also defined cryptocurrency as the "native currency of AI agents."

Coinbase's experiment is interpreted as an attempt to go beyond adopting a simple productivity tool and expand from within a practical structure that links AI agents to cryptocurrency payments and trading. The pace of growth in agent numbers and the level of actual work and transaction automation are expected to emerge as key points to watch.