Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ameo.agiwithai.com/llms.txt

Use this file to discover all available pages before exploring further.

Verifiable cognition means an autonomous agent’s reasoning and authorization steps produce independently checkable artifacts — not just a narrative after the fact.

​
Beyond explainable AI

Explainable AI (XAI) often means post-hoc summaries: “the model chose X because Y.” That prose is not evidence. Verifiable cognition requires structured, immutable commitments at each stage:
  1. Observation snapshot — balances, gas, market inputs at decision time.
  2. Reasoning trace — hashed plan + optional 0G blob.
  3. Policy proof — predicate results evaluated outside the model.
  4. Settlement proof — tx hash, block, and [DecisionLogged](/glossary#decisionlogged-event) event.
An auditor replays the cycle without trusting the operator’s UI.

​
The four pillars in AMEO

PillarArtifactWhere to verify
ObservationJSON snapshot in JSONL + replay API/app/replay node 1–3
ReasoningrationaleHash, planner outputReplay node 4 + 0G indexer
Policyguardrail_evaluated violations or passReplay node 5
SettlementMantlescan tx + DecisionLoggedReplay nodes 6–8
Empty states are honest — if 0G is not configured, the UI says so instead of showing a placeholder hash.

​
Why it matters for production

Capital allocators need replay, not trust. Verifiable cognition is the minimum bar for autonomous treasuries, agentic wallets, and policy-bound DEX execution on public chains. Next: Agent identity · Proof trail