Investigate

Reason through incidents with greater depth and consistency.

Haven brings evidence, timelines, causal reasoning, failed controls, root cause analysis, corrective actions, and report development into one AI-guided investigation workflow.

Investigators stay in control of every conclusion. Haven organizes evidence, tests reasoning, and surfaces gaps a manual review might miss.

Investigation workspace
Evidence synthesis
14 SOURCES · 96% COMPLETE
Causal pathway
3 THREADS · 1 UNRESOLVED
Control analysis
1 FAILED · 2 DEGRADED
Corrective actions
4 DRAFTED · 1 PENDING OWNER
The investigation problem

Depth and consistency are hard to guarantee at scale.

Complex incidents generate evidence from many sources: witness accounts, documents, photographs, sensor data, procedures, and prior records. Assembling that evidence into a coherent, defensible account takes significant time, and the depth of reasoning can vary widely from one investigator to the next.

Haven gives every investigator the same structured starting point: evidence organized, timelines built, causal threads tested, and gaps surfaced before the report is written.

AES has seen an 80% reduction in RCA investigation labor time through adoption of Haven Safety AI. This allows our teams to spend their valuable time closer to our operations where they are invaluable.

Evaristo Leonardi
Evaristo Leonardi
Chief Operations Officer, AES Utilities
What Investigate does

One workflow, from raw evidence to a defensible report.

Evidence synthesis

Bring witness accounts, documents, images, and records into one organized evidence base.

Timeline development

Reconstruct the event sequence and flag where the timeline is incomplete or contradicted.

Contradiction detection

Surface conflicts between sources so investigators can resolve them before they harden into conclusions.

Multi-threaded causal reasoning

Explore more than one causal pathway rather than stopping at the first plausible explanation.

Control-focused analysis

Examine which controls failed, degraded, or were missing, not just what the immediate cause was.

Corrective action development

Connect proposed actions directly to the causes and failed controls identified.

Workflow

From first evidence to a structured report.

01

Evidence is captured

Witness statements, documents, photographs, and records are brought into one workspace.

02

Haven organizes and cross-references

Evidence is structured into a timeline, and contradictions or gaps are flagged for the investigator.

03

Causal pathways are explored

Haven supports multi-threaded reasoning rather than a single linear cause chain.

04

Controls are examined

Failed, missing, or degraded controls are analyzed alongside the causal findings.

05

Corrective actions are drafted

Actions are tied to specific causes and control failures, with an owner and verification plan.

06

The report is assembled

Evidence, reasoning, and actions are compiled into a structured, defensible investigation record.

Familiar methods, sharpened by AI

Support the methods your organization already trusts.

Haven does not force every organization into a single method. It strengthens the depth and consistency of whichever reasoning approach your standard requires, grounded in the Haven Industry Knowledge Graph.

01
Multi-Threaded RCA

AI-inferred causal pathways for complex events.

Serious incidents rarely trace back to a single cause. Haven identifies and traces multiple causal threads that unfold in parallel, reinforced against historical incident patterns rather than treated as isolated failures.

  • Parallel, AI-inferred causal threads
  • Mapping interactions across conditions, actions, and failures
  • Automated detection of missing contributing factors
  • Timeline alignment using collected evidence
  • Full human-in-the-loop control of the analysis
02
Five Whys

Five Whys with AI guardrails and knowledge graph context.

Each Why is grounded in evidence and anchored to the Industry Knowledge Graph. Haven flags circular reasoning, vague causal statements, and breaks in logic as the analysis progresses.

  • AI-generated suggestions for deeper Why steps
  • Guardrails against repetition and abstraction
  • Links to prior incidents with similar cause patterns
  • Knowledge-graph-aligned progression
  • Additional evidence recommendations
03
Fishbone Analysis

AI-assisted cause classification across standard domains.

Haven categorizes causes under people, process, equipment, environment, materials, and management, then highlights missing evidence and factors that appear in similar incidents.

  • AI-recommended contributing factors by category
  • Automated identification of category gaps
  • Evidence-anchored classification
  • Full human-in-the-loop control over the analysis
Corrective actions

Recommendations grounded in causal reasoning, not generic checklists.

Haven generates corrective action recommendations using causal reasoning, the Industry Knowledge Graph, and patterns from prior incidents. Each recommendation is scored for effort and impact, mapped to the Hierarchy of Controls, and reflects interventions proven effective in similar events.

  • AI-generated corrective actions tied directly to root causes
  • Scoring based on effort and estimated impact
  • Knowledge-graph-aligned control mapping
  • Learning from historical incident outcomes
  • Traceability back to the RCA pathway that produced them

Safety Initiative Matrix

EFFORT VS. IMPACT ANALYSIS
AI RANKED
LOW EFFORT HIGH EFFORT HIGH LOW IMPACT Quick Wins Major Projects Fill In Reconsider
2 Quick Wins2 Major Projects3 Fill In
OPTIMAL: FOCUS TOP-RIGHT
Grounded reasoning

Powered by the Haven Industry Knowledge Graph.

Haven's reasoning is grounded in a knowledge graph built from high-signal industry sources, including regulatory frameworks, incident investigation findings, and recognized safety management standards. This gives Investigate a working understanding of hazards, controls, exposures, and failure modes, rather than relying on generic language patterns alone.

Output and documentation

Investigation records built to withstand scrutiny.

Haven produces comprehensive, consistent outputs that safety teams can use internally or share with regulators, auditors, and insurers. Every output is linked back to source evidence and causal logic.

Multi-threaded causal diagrams

Visual maps of parallel causal pathways, tied to the evidence that supports them.

Five Whys chains

Fully documented Why sequences with evidence citations at each step.

Fishbone diagrams

Category-organized causal factors, generated directly from the investigation.

Corrective action plans

Prioritized actions with effort and impact scoring, ready for assignment.

Executive investigation summary

A concise, defensible account of the incident, its causes, and the response.

Give every investigation the same depth of reasoning.

See how Haven can help your organization investigate more consistently while keeping investigators accountable for the conclusions.