Why AGLedger™
Automated work is crossing trust boundaries — between agents, between agents and enterprise systems, between departments, between companies. When something is disputed, audited, or reconstructed, the usual evidence stops being enough. Application logs are attributable to the system that wrote them. Agent self-reports can be log-shaped fiction. AGLedger produces the record that survives these boundaries.
Reach for AGLedger when
Agents act with real consequences. Your agents commit to customers, modify records in your ERP or CRM, initiate transactions, approve payments, or execute decisions that need to be justifiable later.
Work crosses an accountability boundary. Between agents, between agents and systems, between departments, between organizations. Each party has their own logs; no shared source of truth spans the boundary.
Probabilistic participants are in the chain. LLM agents hallucinate, confabulate, and lose context. Their self-reports can be fabricated without malice. The fix is structural: record intent before the work begins.
Audits, disputes, or regulator questions come later. Reconstructing from logs takes weeks and produces guesses. A signed chain answers in seconds, with cryptographic proof attached.
What AGLedger provides
A protocol. Commitment, delivery, decision, closure. Pre-defined contracts and workflows give agents and systems a shared vocabulary, so neither has to improvise. Works across any LLM, any provider, any organization.
An Agent-friendly API. Native REST as the primary path. TypeScript and Python SDKs on npm and PyPI. A CLI and an MCP server are available. Any participant that speaks HTTP can call it — agents, services, enterprise systems, federated partners.
A signed, append-only vault. Customer-owned PostgreSQL. Your Ed25519 keys. Append-only and tamper-evident. Standard tables, no proprietary formats. Air-gapped ready. We provide the software; you own everything else.
Audit-ready records. Structured as accountability from the moment they are written — not reconstructed after the fact. Point your Audit Agent at the vault and answer any audit question in seconds.
Use cases
Solo agents to multi-agent chains
The same protocol handles one agent serving one user, agents delegating to other agents, and systems assigning work to agents. Every handoff carries its own mandate; the full delegation tree is reconstructible from a single mandate ID.
Federation & cross-company accountability
When work crosses organizational lines, federation extends the chain of custody. Each party keeps their own keys and data — the protocol crosses boundaries, the data doesn't.
Compliance & regulatory evidence
Records are audit-ready the moment they're written. EU AI Act, NIST AI RMF, ISO/IEC 42001 — mapped, tested, exportable. Signed proof instead of reconstructed summaries.
How AGLedger compares
How AGLedger differs from logging, observability platforms, policy gateways, continuous-compliance tools, and manual audit trails — and which questions belong to which tool.