> For the complete documentation index, see [llms.txt](https://functor.gitbook.io/functor-network-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://functor.gitbook.io/functor-network-docs/overview/how-is-functor-enabling-the-onchain-autonomous-world.md). # How is Functor enabling the onchain autonomous world? The onchain autonomous world is the union of **AI** and **crypto** driving **automation** at scale. AI proposes actions, crypto enforces outcomes. For this future to work, **programmable execution** and **self‑custody** must be correct by design across chains. Functor is the **onchain signing infrastructure** that makes this possible, turning AI intent into verifiable onchain execution with one consistent way to authorize, constrain, and recover actions across ecosystems. Without this layer, large‑scale automation fragments into per‑chain keys, inconsistent approvals, and fragile recovery. **Here’s how we get there**: safe state changes with **programmable permissions** and policies, **cross‑chain** authorization by design, a session lifecycle with rotation and native recovery, and composable coordination, for everyone. *** ### A quick intuition2 AI is adaptive and **probabilistic**. Crypto is rule‑bound and **deterministic**. You can fake a captcha, but not an **onchain balance** or a **cryptographic proof**. In a world where AI can make things look fake, crypto makes results real again. *** ### What an agent actually does 1. **Read** data 2. Decide and **act** (change state) 3. New data is **produced**
The critical moment is the state change. This is where security and UX usually break. Functor makes this step safe and consistent. *** ### Safe state changes with session keys and policies Agents and platforms need controlled signing, not master keys. Functor enables **session keys** with **programmable permissions**, enforced onchain. * Scoped access to assets, contracts, methods, and limits such as value, frequency, and time. * Onchain verification checks every action against policy before it executes. * Clear intent and predictable behavior for users, apps, and other agents. ### Cross‑chain by design Policies live in Functor’s signing layer and are readable in a chain‑agnostic way. The same rules apply everywhere. * One authorization root that follows the user and their applications across L2s. * No extra approvals when moving between chains. * Consistent **revocation**, **recovery**, and policy updates from one place. An example of how AI agents work with Functor sessions is as follows:
*** ### Session lifecycle and safety controls * Provision: create a session for an agent, a fleet, or a platform with scoped permissions. * Operate: execute under the same policy across chains, continuously. * Rotate or revoke: update keys, tighten limits, or shut down access globally. Supports **native account recovery** via **key rotation** for wallets and exchanges. ### Composable coordination across many actors Because policies are explicit and verifiable, multiple agents, services, and platforms can coordinate safely. * Predictable behavior enables agent‑to‑agent and service‑to‑service workflows. * Relationship policies can enforce ordering or dependencies across assets and chains, for example “Asset B cannot move until Asset A moves first,” without custom cross‑chain messaging.
*** ### Footnotes[​](https://docs.securitylabs.xyz/what_is_functor/#footnote-label) 2. The 8th point on Balaji's [post](https://balajis.com/p/ai-is-polytheistic-not-monotheistic) about his 10 thoughts on AI being economically useful.