VELLA

VELLA is a system for stateful intelligence that focuses on continuity, reflection, and trust across time. Its materials describe it as a substrate for AI governance, with an emphasis on turning possibility into stable, accountable form and keeping a trace that can be inspected later.

A central part of the system is a pre-execution policy called decision authority. Candidate actions are evaluated against four criteria — scope, intent, authority, and budget — before they are allowed to execute or denied. The page also describes this as deterministic, pre-execution adjudication with evidence-bound decisions and signed proof bundles. Each decision, whether allow or deny, is recorded in a cryptographically signed proof ledger. The ledger uses ED25519 signatures and hash-chained records, and the interface shown includes entries such as file reads, codebase searches, shell commands, and file writes.

VELLA is presented for agentic systems, especially where authority, inspectability, and accountability matter. The wording on the site also refers to agent governance and to multiple possibilities being held in structure until one path is adjudicated and resolved. No pricing, licensing, deployment model, or named integrations are stated on the page. The site identifies Vella Cognitive, LLC in its footer and lists a contact email at [email protected].