From Evidence to Knowledge

ACTA is the knowledge layer of the Ardyn platform. It converts the firehose of operational evidence — AEAs, ATPs, governance decisions, and execution traces — into durable organizational cognition. Not a dashboard. Not a report. A cognitive memory that compounds over time and survives personnel turnover.

How does an organization build knowledge from machine evidence?

An organization runs autonomous agents. Every day, those agents execute thousands of actions, produce thousands of AEAs, and generate hundreds of governance decisions. The question is not whether the evidence exists — it does. The question is: how does the organization convert this firehose of operational evidence into durable knowledge it can act on?

Operational evidence is ephemeral. An AEA records what happened at 10:32 AM. A governance decision records that Axiom blocked a payment at 2:15 PM. But knowledge — the kind that shapes strategy, informs policy changes, and survives personnel turnover — requires synthesis. ACTA answers: how does an organization build cognitive memory from machine evidence?

What is ACTA?

ACTA (Ardyn Cognitive Trust Architecture) is the knowledge layer of the Ardyn platform. It converts operational evidence into durable organizational cognition — not dashboards, not reports, but structured, queryable, causal knowledge about how the organization's autonomous systems behave.

Operational Memory

Think of ACTA as the organization's operational memory. Just as human cognition converts raw sensory input into durable memories that inform future decisions, ACTA converts raw AEA streams, ATP profiles, and governance traces into structured knowledge that persists beyond any single operator or agent run. Currently, ACTA surfaces evidence through 8 MCP tools that expose the ingestion, pattern recognition, and knowledge retrieval pipeline — making organizational cognition queryable, verifiable, and actionable through standard interfaces.

ACTA consumes

  • AEAs — individual action records
  • ATPs — agent trust profiles
  • Axiom governance decisions — ALLOW, BLOCK, ESCALATE, DEFER outcomes
  • AAECP execution traces — full pre-execution to post-execution pipelines
  • Forge authority events — grants, delegations, revocations, quorum decisions

ACTA produces

  • Behavioral patterns — which agents act in which ways, under which conditions
  • Risk correlations — which actions, agents, or contexts most frequently produce escalations or blocks
  • Policy effectiveness metrics — how often do policies fire, how often do they produce the intended outcome
  • Authority utilization — which grants are exercised, which are dormant, which are challenged
  • Trust drift — how agent behavior changes over time, and whether the change is benign or concerning

The key distinction: ACTA is not a dashboard. A dashboard shows you what happened in the last hour. ACTA shows you what the organization has learned — and what it should do differently — based on months of evidence.

Why existing approaches are insufficient

Dashboards and monitoring

Grafana, Datadog, and similar tools show real-time metrics. They answer "what is happening now?" They do not answer "what pattern has emerged over the last six months?" Dashboards are operational tools. ACTA is a cognitive tool.

Business intelligence platforms

Tableau, Looker, and Power BI can analyze data, but they require the data to be structured, cleaned, and loaded into a warehouse. They do not understand Ardyn primitives — AEAs, ATPs, governance traces — and cannot produce causal analysis without a data engineering team building the pipeline.

Post-hoc audits

An auditor reviews a sample of actions after the fact. This catches problems that already happened. It does not detect patterns early, does not inform policy changes in real time, and does not scale to the volume of actions autonomous systems produce.

Institutional memory in people

The engineer who built the agent knows how it behaves. When that engineer leaves, the knowledge leaves with them. ACTA makes organizational knowledge machine-readable and durable — it survives personnel changes.

ACTA provides a cognitive layer that is natively integrated with Ardyn evidence: it understands AEAs, governance traces, and trust profiles without translation layers, and it produces structured knowledge that can feed directly back into policy and authority configuration.

How ACTA builds organizational cognition

ACTA operates through a continuous cognition loop:

4.1 Ingestion

Evidence flows into ACTA through the Ardyn pipeline:

AEAs ──────────────┐
ATPs ──────────────┤
Axiom decisions ───┼──► ACTA Ingestion ──► Cognitive Store
AAECP traces ──────┤
Forge events ──────┘

Each artifact is validated before ingestion: content digest verified, signature checked, chain position confirmed. ACTA only ingests verified evidence.

4.2 Pattern Recognition

ACTA applies computational analysis to the ingested evidence:

AnalysisQuestion
Behavioral clusteringWhich agents exhibit similar action patterns?
Risk correlationWhat conditions precede escalations or blocks?
Policy effectivenessDoes a policy rule produce the intended outcome?
Authority analysisWhich grants are exercised, which are dormant?
Trust drift detectionIs an agent's behavior changing over time?

4.3 Knowledge Production

The output of analysis is structured knowledge:

  • Cognitive reports — machine-readable documents that describe patterns, correlations, and anomalies with evidence citations (AEA IDs, ATP fields, governance trace hashes)
  • Policy recommendations — suggested policy changes based on observed behavior
  • Risk alerts — proactive notifications when behavior patterns deviate from historical norms

4.4 Feedback Loop

The knowledge produced by ACTA feeds back into the system:

  • Policy adjustment. An organization updates its Axiom policy based on ACTA's policy effectiveness analysis.
  • Authority reconfiguration. An organization revokes dormant authority grants or adds constraints based on ACTA's authority utilization analysis.
  • Agent reconfiguration. An agent's tool set or constraints are adjusted based on ACTA's behavioral analysis.
┌──────────────────────────────────────────────────────────────────┐
│                    ACTA Cognition Loop                          │
│                                                                 │
│  ┌──────────┐     ┌──────────────┐     ┌──────────────────┐    │
│  │ Evidence │────►│ Pattern      │────►│ Knowledge        │    │
│  │ Ingestion│     │ Recognition  │     │ Production       │    │
│  └──────────┘     └──────────────┘     └────────┬─────────┘    │
│                                                  │              │
│                                                  ▼              │
│  ┌──────────┐     ┌──────────────┐     ┌──────────────────┐    │
│  │ System   │◄────│ Policy /     │◄────│ Feedback to      │    │
│  │ Behavior │     │ Authority    │     │ Organization     │    │
│  │          │     │ Adjustment   │     │                  │    │
│  └──────────┘     └──────────────┘     └──────────────────┘    │
│                                                                 │
└──────────────────────────────────────────────────────────────────┘

ACTA in practice

A large financial institution runs 200 autonomous agents across five departments: trading, compliance, settlement, risk, and reporting. Over 18 months, the agents have produced 1.2 million AEAs, 8,000 governance decisions, and 200 ATPs.

Operational question: "Are our trading agents becoming more or less aggressive over time?"

Without ACTA: a compliance officer manually reviews a sample of trades, forms a subjective impression, and writes a report. The process takes two weeks and is not reproducible.

With ACTA:

1. ACTA ingests all trading AEAs over the trailing 18 months.

2. Pattern recognition identifies that the average trade size has increased by 12% and the frequency of trades above the $1M escalation threshold has increased by 40%.

3. ACTA produces a cognitive report: "Trading agents show a statistically significant increase in trade size and frequency of high-value trades over the trailing 18 months. The escalation rate has risen from 2.1% to 3.8%, concentrated in the equity derivatives desk."

4. The report cites specific AEA IDs, ATP fields, and governance traces. Every claim is evidence-backed.

5. The institution's risk committee reviews the report and decides to: lower the escalation threshold for equity derivatives from $2M to $1.5M, and require quorum authorization for equity derivative trades above $500K.

6. The policy changes are deployed to Axiom. ACTA begins monitoring the impact.

Six months later: ACTA reports that the escalation rate has dropped to 1.4%, trade sizes have stabilized, and no unauthorized trades have been detected. The cognitive loop has closed.

What ACTA does not do

Make governance decisions

ACTA produces knowledge and recommendations. It does not ALLOW, BLOCK, or ESCALATE actions. That is Axiom's domain.

Execute actions

ACTA analyzes evidence. It does not execute or authorize actions. That is AAECP's domain.

Replace human judgment

ACTA surfaces patterns and suggests actions. A human decides whether to act on the recommendations. ACTA is a cognitive augmentation tool, not an autonomous decision-maker.

Provide real-time monitoring

ACTA is a knowledge layer, not an operational monitoring tool. It answers longitudinal questions, not "what is happening right now?"

Predict future behavior with certainty

ACTA detects trends and correlations. It does not predict what an agent will do next Tuesday at 3:00 PM. It identifies patterns that inform human judgment.

How ACTA relates to other layers

DirectionLayerRole in ACTA
UpstreamAEAIndividual action records that form the raw material of cognition.
UpstreamATPAgent trust profiles that provide longitudinal behavior context.
UpstreamAxiomGovernance decisions that reveal policy effectiveness and risk patterns.
UpstreamAAECPExecution traces that provide the full authorization-to-closure picture.
UpstreamForgeAuthority events that reveal how organizational authority is exercised.
DownstreamAxiomACTA's policy effectiveness analysis informs policy adjustments.
DownstreamForgeACTA's authority utilization analysis informs authority grant configuration.
DownstreamHuman decision-makersACTA's cognitive reports and risk alerts support organizational governance.
LateralVerificationACTA's ingestion path validates every artifact before analysis. ACTA only reasons about verified evidence.

How ACTA knowledge claims are independently verifiable

Evidence traceability

Every claim in an ACTA cognitive report is backed by cited AEA IDs, ATP fields, or governance trace hashes. A verifier can retrieve the underlying evidence and confirm that the claim is supported.

Analytical reproducibility

ACTA's analytical methods are deterministic. Given the same set of AEAs, the same behavioral clustering algorithm will produce the same clusters. An auditor can re-run ACTA's analysis and confirm the results.

Causal chain integrity

When ACTA recommends a policy change, the full causal chain — evidence → pattern → analysis → recommendation → policy change → outcome — is recorded in the governance trace. A verifier can walk the entire chain and confirm that each step is supported by evidence.

Current implementation status

LIVE IMPLEMENTED PLANNED ACTA is the knowledge layer in the Ardyn architecture — implemented in Kylewilson04/acta @ 494959f. MCP tools: Kylewilson04/acta-mcp @ 884ccb2. Core evidence surfacing is LIVE. The 8 MCP tools are registered and operational, exposing the ingestion, pattern recognition, and knowledge retrieval pipeline. Operational memory is IMPLEMENTED via these tools. Full organizational cognition (behavioral clustering, risk correlation, policy effectiveness, trust drift detection) is PARTIAL — pattern surfacing works, but advanced automated analysis is defined and awaiting implementation.

Evidence ingestion pipeline LIVE

Operational through the existing AEA, verification, and governance infrastructure. AEAs are produced and verifiable today.

MCP tools (8 tools) LIVE

ACTA surfaces evidence through 8 MCP tools registered in api/mcp-tool-registry-v1.json. These expose ingestion, pattern recognition, and knowledge retrieval — making organizational cognition queryable through standard MCP interfaces. (CLAIM-LEDGER #15)

Operational memory IMPLEMENTED

The 8 MCP tools provide queryable access to evidence streams, agent behavior patterns, and governance traces — implementing ACTA's operational memory function. Evidence is surfaced and retrievable through standard interfaces.

Organizational cognition PARTIAL

Pattern recognition and knowledge retrieval are live through MCP tools. Advanced automated analysis — behavioral clustering, risk correlation, policy effectiveness metrics, and trust drift detection — is architecturally defined but not yet implemented as automated cognitive reports. (See CLAIM-LEDGER.)

ACTA represents the final layer of the Ardyn platform: the point at which evidence becomes knowledge, and knowledge feeds back into governance. It is the layer that closes the cognition loop.

Continue learning

ACTA is the knowledge layer that sits atop the Ardyn trust plane. Each linked layer feeds into or consumes ACTA's cognitive output.