Turnkey Unveils Managed Secrets to Secure AI Agents and Agentic Finance Workflows

Turnkey has launched Managed Secrets, a new infrastructure tool designed to secure sensitive credentials for AI agents and applications. By extending its secure enclave technology beyond private keys to API keys and passwords, the company aims to solve the security bottleneck where automated agents require access to sensitive data without exposing it to potential compromise.

What was announced

Turnkey Managed Secrets is an expansion of the firm’s existing private key management infrastructure. While the platform previously focused on securing onchain transactions, the new product applies the same secure enclave and policy engine framework to a broader range of sensitive values, including API keys, passwords, payment details, keyshares, and recovery keys.

The system is built to address the specific security risks posed by AI agents. Traditionally, credentials like exchange API keys are stored in environment variables or configuration files, which are vulnerable if the underlying infrastructure is breached. Managed Secrets stores these values inside secure enclaves and only releases them, encrypted, when specific programmatic policies are met.

The workflow involves four distinct stages: an application imports a secret using encryption; the enclave re-encrypts it for storage; the enclave validates requests against user-defined policies; and the secret is delivered encrypted to the intended recipient. This recipient can be the requesting agent or a separate service. Users can configure access patterns ranging from direct autonomous access to multi-agent approval or mandatory human intervention via the Turnkey dashboard. Notably, requests for multiple secrets are handled as all-or-nothing transactions; if any single secret in a batch fails a policy check, the enclave releases none of them.

The companies involved

Turnkey is a developer-focused infrastructure provider specializing in secure key management and onchain operations. The company provides the underlying security architecture that allows developers to build and scale crypto-native applications without managing the complexities of hardware security modules or manual signing processes. Its platform is built on secure enclaves, which isolate sensitive cryptographic operations from the rest of the application environment to mitigate the risk of credential theft.

The firm has established itself as a critical layer in the decentralized finance and digital asset ecosystem, securing billions of dollars in transaction volume. By focusing on programmatic access and policy-driven security, Turnkey caters to a market of fintech firms and crypto platforms that require high-velocity transaction signing and automated asset management. The company’s expansion into general secret management reflects a broader shift in the industry toward securing the entire lifecycle of an automated agent, moving beyond simple wallet interactions to encompass the full stack of credentials required for modern financial applications.

What FF News has reported before

FF News has followed Turnkey’s expansion into broader treasury and asset management tools. In October 2026, the company launched Turnkey Launches Console to Streamline Onchain Treasury and Asset Operations, which introduced a centralized interface for managing security controls. This followed the September 2026 release of Turnkey Launches Swaps and Earn to Monetize Embedded Wallets with In-App Trading, a move that enabled developers to integrate trading features directly into their applications. Turnkey’s infrastructure also underpins broader industry moves, such as when Modern Treasury Unveils Non-Custodial Stablecoin Wallets to bridge fiat and crypto rails, highlighting the demand for secure, non-custodial wallet solutions in the enterprise space.

What this means

This move highlights a critical pivot in the fintech security landscape: the transition from human-centric security to agent-centric security. As AI agents take on more autonomous financial tasks, the industry faces a "least privilege" dilemma where giving an agent a key often means giving it the "keys to the kingdom." By decoupling the use of a secret from the visibility of that secret, the market is moving toward a model where credentials are never held in plaintext by the systems that use them. This puts pressure on traditional vault providers who have struggled to adapt to the high-frequency, programmatic demands of autonomous agents.