Evidence-First Infrastructure

Forensic reconstruction for
AI-assisted decisions

Turn existing logs into evidence-first timelines and auditor-friendly incident narratives—without overclaiming.

Evidence-first
Works with what you have
Explicit uncertainty
Marks gaps as unknown
Human oversight captured
Approvals & interventions
Demo Data

Decision Path Reconstruction

Observe how AI and human decision events are captured and reconstructed—not judged or scored.

AI Event ObservedTreasury Bot 04
13:00:00

Agent initiated $50,000 transfer request to external account.

KNOWN
sha256:8f2a3...5b6c
trace-2026-0...-001
Policy Reference AttachedPolicy Engine
13:00:12

Transfer amount flagged against automated limit threshold of $10,000.

KNOWN
sha256:7c1b2...4c5d
Human Intervention RecordedSenior Auditor
13:02:45

Manual override recorded. Context: Emergency Liquidity Event.

KNOWN
sha256:5d9e8...6f5a
Decision Path ReconstructedReconstruction Engine
13:03:00

Accountability path reconstructed for review. Chain integrity preserved.

INFERRED
sha256:a1b2c...c5d6
Demonstration Data Only

This timeline illustrates forensic reconstruction capability. It does not make compliance determinations or assign liability.

What We Do

Three steps to turn scattered evidence into auditor-ready timelines.

1

Ingest what you have

Upload structured or messy logs—AI traces, tickets, approvals, screenshots, policy PDFs. No SDK required to start.

2

Reconstruct decision paths

We build a forensic timeline showing AI actions, human interventions, and system events. Every entry is marked Known/Inferred/Unknown.

3

Produce evidence bundles

Generate auditor-friendly timelines and narratives with cryptographic hashes. Analysis support, not compliance verdicts.

Built for Risk, Legal, and Compliance Teams

When an AI-assisted decision is questioned, you need to reconstruct what happened using incomplete logs. AegisTrace is purpose-built for that challenge.

Chief Risk Officers

Build defensible evidence chains for AI-assisted decisions when incidents are questioned.

General Counsel & Legal

Reconstruct decision timelines with timestamped evidence for regulatory inquiries and litigation support.

Compliance Officers

Prepare auditor-friendly documentation showing what happened, what was known, and where gaps exist.

Audit & Security Teams

Investigate AI decision incidents with forensic-grade reconstruction and explicit uncertainty markers.

What We Are — What We Are Not

Clear boundaries matter. We provide evidence infrastructure, not compliance verdicts.

What We Are

  • Evidence capture and timestamping infrastructure
  • Forensic decision path reconstruction
  • Analysis support for compliance and legal teams
  • Transparent about gaps (Known/Inferred/Unknown)
  • Starting point: upload what you already have

What We Are Not

  • Compliance determination or violation detection
  • Risk scoring or pass/fail certification
  • Prevention or real-time blocking system
  • Guarantee of safety or regulatory approval
  • Requiring perfect logs or full AI instrumentation

Explicit Uncertainty

We surface what is known, what can be inferred, and what remains unknown. No guesswork. No false confidence.

Known

Directly observed in evidence

Example

Agent ID, timestamp, logged action

Inferred

Derived deterministically

Example

Time divergence, handoff detection, sequence analysis

Unknown

Missing evidence and explicitly flagged

Example

Human rationale not captured, partial logs

This transparency builds trust with regulators and internal reviewers who need to understand the limits of what can be reconstructed.

Start With What You Have

We don’t require perfect logging. We work with partial data and mark gaps as unknown.

AI logs
JSON/NDJSON/exported traces
Tickets & notes
Approvals, screenshots, manual records
Policy PDFs
Existing documentation
Spreadsheets
CSV exports and data tables

No SDK or agent required

Upload what you have. We reconstruct the decision path and explicitly mark missing evidence.

Forensic Event Feed

Every AI and human decision event captured with cryptographic proof. Immutable audit trail for regulatory investigation and review.

AI Event
Reconstruction Complete
evt_2026_001_au_nsw
Document classification performed
AU-NSW
13/01/2026, 21:15:42 UTC
Agent ID
agent_legal_counsel_01
Data Provenance
CUSTOMER
Human Action
No
Regulatory Scope
APRA CPS 230
Privacy Act 1988
Policy References Available
APRA-CPS-230-3.1
PRIVACY-1988-5.2
Chain Hash (Cryptographic Proof)
sha256:a7f5c8d9e2b4f1a8c6d3e9b2f4a1c8d6e3b9f2a4c1d8e6b3f9a2c4d1e8b6f3a9
Human Intervention
Handoff Detected
evt_2026_002_eu_gdpr
Automated decision on personal data with human review
EU-GDPR
13/01/2026, 20:58:17 UTC
Agent ID
agent_data_handler_03
Data Provenance
THIRD_PARTY
Human Action
Yes
Regulatory Scope
GDPR Art. 22
EU AI Act
Policy References Available
GDPR-ART-22
EU-AI-ACT-15
Chain Hash (Cryptographic Proof)
sha256:d1h8f2c5e7b9a3d1f8c6e4b2a9d7f5c3e1b8d6a4f2c9e7b5d3a1f8c6e4b2d9a7
Path Reconstructed
Reconstruction Complete
evt_2026_003_us_ca
Consumer data access request processed
US-CA
13/01/2026, 19:42:33 UTC
Agent ID
agent_risk_assessor_02
Data Provenance
HYBRID
Human Action
No
Regulatory Scope
CCPA
CPRA
Policy References Available
CCPA-1798-100
CPRA-1798-105
Chain Hash (Cryptographic Proof)
sha256:f3j1h4e8c2b6d9f3a7e1c5b8d2f6a9c4e7b1d5f8a3c6e2b9d7f4a1c8e6d3b5f9
Scenario Library

Explore linked scenarios and incident reconstructions

Browse cross-domain scenarios that demonstrate decision timelines, evidentiary artifacts, and explicitly marked uncertainty—without breaking the narrative flow.

Frequently Asked Questions

Common questions about how AegisTrace works and what it delivers.

Decision Event Schema

Illustrative data structure for evidentiary capture. Demonstrates how events are cryptographically recorded for review.

Illustrative Event Object
This demonstrates the type of data captured in forensic reconstruction. Chain integrity maintained via SHA-256 hashing.
{
  "event_id": "evt_2026_001_au_nsw",
  "timestamp": "2026-01-13T21:15:42.123Z",
  "agent_id": "agent_legal_counsel_01",
  "jurisdiction": "AU-NSW",
  "regulatory_scope": [
    "APRA CPS 230",
    "Privacy Act 1988"
  ],
  "event_type": "AI_EVENT_OBSERVED",
  "decision_stage": "RECONSTRUCTION_COMPLETE",
  "chain_hash": "sha256:a7f5c8d9e2b4f1a8c6d3e9b2f4a1c8d6e3b9f2a4c1d8e6b3f9a2c4d1e8b6f3a9",
  "previous_hash": "sha256:b8f6d9e1a3c5f2b9d7e4f1a9c7d4e2b5f3a1c9d7e4b2f5a3c1d9e7b4f2a5c3d1",
  "data_provenance": "CUSTOMER",
  "hashed_reasoning": "sha256:c9g7e1b4d6f3a2c9e8b5f1d7a4c2e9b6f3d1a8c7e5b2f9d4a1c8e6b3f2d9a5c7",
  "observed_action": "Document classification performed",
  "human_action_present": false,
  "policy_reference_ids": [
    "APRA-CPS-230-3.1",
    "PRIVACY-1988-5.2"
  ],
  "policy_context_available": true
}
Immutable Tracing
chain_hash

Cryptographic fingerprint linking events in sequence

previous_hash

Reference to prior event creating unbroken chain

Event Observation
event_type

Category of observed event (AI action, human intervention, etc.)

decision_stage

Current reconstruction phase of the decision path

Forensic Context
hashed_reasoning

SHA-256 hash preserving AI deliberation for later review

observed_action

Human-readable description of action recorded

Regulatory Mapping
data_provenance

Tracks origin of data (customer, third-party, hybrid)

policy_reference_ids

Links to applicable policy contexts for investigative review