Ledger Launches Agent Stack to Keep Humans in Control of AI-Driven Crypto Transactions

Ledger has rolled out an open source toolkit called Agent Stack, designed to let AI agents work with crypto wallets while making sure humans still approve any sensitive action before it happens.

How the System Works

Under this setup, AI bots can check balances, suggest trades, prepare swaps, and draft transaction proposals on their own. But none of these actions can actually go through without a person physically confirming them on a Ledger hardware device. This puts Ledger in a more cautious camp compared to some other wallet providers that already let bots execute actions automatically within preset rules.

Why Ledger Is Taking a Cautious Approach

A company representative explained that relying only on software firewalls, AI guardrails, and input filtering isn’t enough anymore as attacks targeting AI agents grow more sophisticated. The representative pointed out that even large companies like Starbucks and Meta have pulled back on AI agent deployments due to privacy concerns and issues with agents acting unpredictably.

Ledger also cited internal findings showing that 26.1% of all agent skills carry at least one security vulnerability. At the same time, human error remains responsible for roughly 60% of security breaches, which is part of why the company is betting on a hybrid model that keeps both AI and human judgment in the loop, rather than removing people from the process entirely.

Beyond Crypto: Securing API Keys and Agent Credentials

Because Ledger devices already support the OpenPGP encryption standard, developers can also use their hardware wallets to encrypt API keys and AI agent credentials, extending the tool’s usefulness well beyond crypto trading. Ledger devices can additionally function similarly to YubiKey security tools, letting users log into services like GitHub, npm, 1Password, and Discord using the same hardware.

Part of a Broader Industry Shift

Thursday’s announcement is part of Ledger’s broader 2026 AI roadmap, which includes plans for future tools covering agent identity verification and “proof of human” confirmation systems. Founded in Paris in 2014, Ledger is reportedly exploring a public listing that could value the company above $4 billion.