Coinbase is testing internal AI agents in Slack and email as the company deepens its push into workplace AI and agent-driven crypto infrastructure. | Credit: CCN.
Share
Key Takeaways
Coinbase is testing AI agents that appear in Slack and email and interact with staff like coworkers, CEO Brian Armstrong said on April 20.
The first two agents are modeled after Coinbase co-founder Fred Ehrsam and former CTO Balaji Srinivasan and are meant to provide employees with high-level feedback.
The experiment aligns with Coinbase’s broader AI push, which already includes internal enterprise agents, AgentKit tooling, and agent-focused wallet infrastructure.
Coinbase is testing workplace AI agents inside Slack and email as the company pushes further into agent-based software across its internal operations and developer stack.
CEO Brian Armstrong said the agents show up “just like any human teammate,” with the first two modeled after former Coinbase leaders Fred Ehrsam and Balaji Srinivasan.
Armstrong said the agents are being introduced to give employees high-level input inside everyday communication channels, and suggested Coinbase will make it easier for teams to create more of them over time.
Sponsored
Disclosure
We sometimes use affiliate links in our content, when clicking on those we might receive a commission at no extra cost to you. By using this website you agree to our terms and conditions and privacy policy.
The Slack and email rollout builds on work Coinbase had already described months earlier.
In a December 2025 engineering post, the company said it formed an “Agentic AI Tiger Team” to build enterprise AI agents and establish internal standards for trust, scale, and repeatability.
Coinbase said those efforts focused on institutional support, onramp onboarding, and listing legal review.
The rollout fits a broader internal AI strategy at Coinbase, where the company has already described agents as tools for real workflows across teams, supported by auditability and monitoring.
From Internal Tools to Onchain Agents
Coinbase is not limiting its AI push to internal workflows.
The company has also been building tools for developers who want AI agents to act on-chain.
Its AgentKit product lets developers give agents access to blockchain actions such as transfers, swaps, token launches, and smart contract interactions.
In February, Coinbase introduced Agentic Wallets, a wallet system designed specifically for AI agents.
The company said the product supports autonomous spending, trading, and API payments, with non-custodial wallets secured in trusted execution environments and connected to Coinbase’s developer stack for authentication, telemetry, and security monitoring.
Coinbase echoed that vision in an April 1 developer newsletter, where it described a future in which AI agents could move across the internet with wallets of their own and transact using stablecoins, tokens, and traditional payment methods.
A Broader AI Strategy Is Taking Shape
The latest test suggests Coinbase is moving toward a model where AI agents appear in both workplace communication and on-chain financial activity.
Inside the company, that could mean agents taking on more advisory, support, or operational roles through the same channels employees already use every day.
Outside the company, Coinbase is also building the infrastructure that those agents could use to transact on-chain.
The rollout is still early. Armstrong said Coinbase is starting with two agents, which suggests the deployment remains limited for now.
But the broader direction is becoming clearer: Coinbase is testing AI agents inside the workplace while building tools for agents to operate on-chain.
