Features / Hazards
Hazards — the leading-indicator surface, before incidents.
Photo + GPS hazard capture from any device, mapped at site level, scored on a configurable risk matrix, classified through the hierarchy of controls, tracked Active → Mitigating → Resolved with SLA-driven corrective-action gates. Job Safety Analysis runs the same discipline pre-task. Twenty-two universal templates seeded at provisioning plus thirty-one pack-tagged categories across Manufacturing, Construction, Mining, and Logistics. Root-cause investigation — 5-Whys, Fishbone, Cause Mapping — attaches to hazards as a first-class subject, with a curated Generic Causes library, AI suggestions anchored to it, per-investigation report PDF, knowledge-base search across every past investigation, and corrective-action effectiveness tracking. SDS uploads run through AI extraction that proposes candidate hazards with grounded citations back to the source PDF. When a hazard becomes an event, the realized-incident backlink closes the loop both ways.
The populated, risk-scored hazard register through stand-alone hazard capture — the running product against Apex Manufacturing demo data. On-screen captions narrate each step.
See it in action
Walk-throughs against live demo data.
Short, captioned clips of the running product against the Apex Manufacturing demo tenant. Click any to play.
What's in it
The capability surface.
Hazard capture + the map view
- Photo + GPS hazard reporting from any authenticated device.
- An interactive site map plots every open hazard per site.
- Point / Circle / Polygon location-shape support on the record.
- Standalone hazard recording — no parent incident required.
- Hazards surfaced from incident investigations carry the originating-incident backlink for traceability.
Configurable risk matrix
- Configurable risk matrix — relabel axis levels, edit per-cell scores, and set tier-band thresholds in the admin UI; the model supports alternate grid shapes (3×3 / 4×4 / 5×5), configured at setup.
- Ships with the SE default: an asymmetric ALARP-aligned 5×5 Likelihood × Consequence grid, 1–25 scale, four tiers (1–4 Low · 5–9 Medium · 10–16 High · 17–25 Extreme).
- Scores are persisted at assessment time so historical assessments survive a later matrix reconfiguration.
- Interactive grid picker with visual tier colour-coding + keyboard accessibility.
Hierarchy of controls + effectiveness review
- Five-level classification — Eliminate → Substitute → Engineering → Administrative → PPE.
- Each control has its own forward-only lifecycle (Planned → InPlace → Verified → Removed), independent of the parent hazard.
- Per-control effectiveness rating on a review cycle — "is this control actually working?" is a tracked answer, not an assumption, with the review date and outcome on the record.
- Verification timestamping + audit stamps on every control.
- Kind-coloured badges for the control hierarchy + status-coloured badges for the lifecycle state.
Hazard lifecycle + corrective-action gating
- Active → Mitigating → Resolved status states with a tracked lifecycle.
- SLA-driven escalation on Active hazards past their threshold.
- Resolved transition is gated — closure requires all open corrective actions to be done.
- Resolved hazards excluded from active pick-lists to prevent contradictory records.
- Auto-generated corrective-action tasks with owner assignment, due-date scheduling, and discussion threads.
Hazard categories — universal + industry-pack
- 22 universal categories seeded at provisioning — Mechanical, Electrical, Pressurised Systems, Radiation, Noise, Vibration, Temperature Extremes, Chemical, Biological, Fire, Explosion, Hot Work, Working at Heights, Slip/Trip/Fall, Confined Space, Lone Working, Manual Handling, Lifting Equipment, Vehicle/Mobile Equipment, Ergonomic, Psychosocial, Workplace Violence/Security.
- Multi-category support per hazard.
- Manufacturing pack — 10 categories: Lockout-Tagout, Machine Guarding, Industrial Robotics, Pressure Vessel, Process Safety, Hot Surface, Welding Fume, Combustible Dust, Conveyor Pinch Point, Forklift / PIT.
- Construction pack — 10 categories: Falls from Heights, Excavation Collapse, Crane / Lifting, Falling Objects, Scaffolding, Power Tools, Heavy Equipment, Trenching, Concrete Pour, Demolition.
- Mining pack — 5 categories: Ground Stability, Blasting, Mine Gas, Underground Ventilation, Heavy Mining Equipment.
- Logistics pack — 6 categories: PIT Safety, Dock Operations, Cold Storage, Loading Bay, Pallet Handling, Conveyor System.
Root-cause investigation — six methodologies + a knowledge base
- The same investigation attaches to both Hazards and Incidents — investigate a near-miss preventively, not only after the recordable event.
- Six methodologies, selectable per investigation — 5-Whys (linear cause chain), Fishbone / Ishikawa (six-category cause analysis), Cause Mapping (parallel evidence-verifiable effect-to-root chains), FMEA (Failure Mode and Effects Analysis with Severity / Occurrence / Detection and live RPN), Bow-Tie (threats → top event → consequences with preventive and mitigative barriers rated by effectiveness), and Apollo / RealityCharting (recursive cause tree with Action and Condition pairing).
- Visual cause-tree diagrams for the three structured methodologies — Cause Mapping, Bow-Tie, and Apollo each render as a read-only diagram alongside the editor, so investigators see the shape of the story at a glance and reviewers can scan a complex tree in seconds.
- Curated Generic Causes library — a catalogue of common root causes across nine cause families with industry-pack variants; the AI suggestions are anchored to it, so they speak your organisation's own cause vocabulary, and tenants extend it without losing the seeded baseline.
- Per-investigation report PDF — subject, methodology working, root-cause conclusion, recommendations, and reopen history, with a tamper-evident content fingerprint.
- Knowledge-base search across every past investigation — "have we seen this before, and what did we conclude?" answered in seconds, faceted by methodology, subject, date, and generic cause.
- Corrective-action effectiveness tracking — each recommendation that becomes an action carries an effectiveness check after it closes; the investigation rolls the outcomes up so a fix that didn't hold is visible, not assumed.
- An "RCA Recommended" banner surfaces automatically on Extreme-tier hazards; concluding an investigation pre-populates the corrective-action tasks from its recommendations.
SDS-driven candidate hazard extraction
- Uploading a Safety Data Sheet triggers AI extraction, producing hazard candidates for review.
- Each candidate carries — extracted name, suggested category codes (resolved against the tenant's catalog), confidence score, grounded-indicator flag, citation back to source PDF page + paragraph.
- Approving a candidate creates the hazard with the source citation preserved for evidentiary traceability.
- Reject captures the reason for audit.
- Cross-chunk dedupe via title similarity + page distance + category-code union.
- Per-tenant extraction (no cross-tenant pattern learning at launch — see "What's evolving" below).
AI-assisted suggestions (advisory only)
- Next-step suggestions across the 5-Whys, Fishbone, and Cause Mapping methodologies — click-to-fill chips drawn from the tenant's own historical investigations and anchored to the curated Generic Causes library, so suggestions speak the organisation's cause vocabulary.
- Fishbone-cause and cause-map-node suggestions — per-category / per-node click-to-fill chips from tenant historical data.
- Three-tier confidence scoring (high / medium / low); low-confidence suggestions render subdued.
- Always click-to-accept — suggestions never auto-fill, never write to a record without explicit user OK.
- Graceful degrade — AI suggestions return empty (not an error) when the AI service is unreachable.
Realized-incident backlink
- On any incident, a multi-select captures which hazards the event was realized from.
- Active-only filter excludes Resolved hazards from the picker (prevents accidental link to a closed record).
- Reverse panel on the Hazard edit page — read-only "incidents that realized this hazard" list.
- Origin distinction — for hazards surfaced from an incident investigation, the originating-incident reference is preserved on the hazard.
Ergonomics assessments — structured MSD-risk capture
- Record structured musculoskeletal-disorder risk assessments using the industry-standard methods — REBA (Rapid Entire Body Assessment), RULA (Rapid Upper Limb Assessment), and the NIOSH Lifting Equation, with an Other catch-all for tenant-specific approaches.
- Capture per-assessment context — assessor, subject (a specific worker or a workstation-scope analysis), task description, score, score interpretation, body regions implicated (neck, shoulder, back, arm, wrist, leg, knee), recommended interventions, and optional follow-up date.
- Site-scoped record alongside the rest of the platform — assessments land in the same audit trail and corrective-action workflow as everything else.
- Without motion-capture AI — gets you to the structured-record yes for the "do you do ergonomics?" eval question without the heavy specialist tooling. If you already use motion-capture analysis, pair it alongside.
IndustryPacks — additive content composition
- Four industry packs — Manufacturing, Construction, Mining, Logistics.
- Tenants select packs in tenant settings; multi-pack tenants see the union of all selected packs' content.
- Seeder reads tenant packs once at start + skips items whose pack tag isn't in the tenant's set.
- Re-seed hook fires when the tenant modifies pack selection; customisations are preserved on pack removal (items stay via upsert-by-Id).
Representative workflows
What this looks like in practice.
Site walk → hazard pinned + scored + controlled → corrective actions land + close
A site walk surfaces a slip-hazard at the loading bay. The walker captures a photo + GPS pin from a tablet, picks Slip / Trip / Fall + Dock Operations (Logistics tenant) from the multi-select, scores Likelihood 4 × Consequence 3 = High, submits. Hierarchy-of-controls planning opens — Engineering (drainage / surface texture), Administrative (housekeeping cadence), PPE (footwear standard). Each control is its own record with its own lifecycle and verification stamp. Auto-generated corrective-action tasks land on ops + facilities + EHS with SLA-derived due dates. The hazard sits in Mitigating until every task closes, then the Resolved transition unlocks. No spreadsheets, no exports.
An SDS upload becomes a candidate hazard with grounded citations
The procurement team uploads a Safety Data Sheet for a newly-onboarded solvent. SE reads the GHS-structured PDF and produces a hazard candidate with the suggested name, the matched category codes against the tenant's catalog (Chemical + Flammable + Hot Work-adjacent), a confidence score, and a citation to the GHS Section 2 paragraph that backed the suggestion. The Site Safety Officer reviews the candidate, edits the name, adds a Health-Hazard category, and approves. The new hazard lands with that citation preserved. If a future inspector asks where this hazard came from, the answer is the page and paragraph of the SDS — not "the safety officer added it last quarter."
An Extreme-tier hazard triggers RCA → corrective actions → realized-incident prevented
A near-miss surfaces a hazard at an underground equipment intersection scored Likelihood 4 × Consequence 5 = Extreme (RiskScore 20). The "RCA Recommended" banner appears on the hazard edit page. The HSE manager opens a 5-Whys session against the hazard as the subject — preventative, not post-event. Three why-step suggestions are tenant-historical (one high-confidence from a sister-site event, two medium); the manager accepts one and edits two. The session concludes with three recommendations; three corrective-action tasks pre-populate and land on owners. Six weeks later, a separate event at a nearby intersection links itself via the realized-from-hazard picker — the audit trail shows the RCA conclusion + corrective actions + realized incident. A leading-indicator chain that closed before it became a fatality.
How this is different
What sets the hazards model apart.
Most EHS platforms ship hazard registers as a separate module from incident management, with a category list, a risk matrix, and a checklist. SE built the hazards register as a peer to incident management — the same root-cause investigation, the same corrective-action engine, the same audit trail, the same connected record. The differences that follow are direct consequences of that decision.
RCA on hazards is a first-class workflow, not "if you have time"
Most platforms restrict RCA to recordable incidents — the high-stakes events. SE attaches all six methodologies to hazards as a first-class subject, with the "RCA Recommended" banner surfacing automatically on Extreme-tier hazards. The point is preventive: running RCA on a near-miss does the work an RCA on the eventual recordable would have done, without the body. Conclusion auto-populates corrective actions; the audit trail is the same shape.
Hierarchy of controls is a structure, not a label
Many platforms record "what controls are in place" as a free-text or checkbox field on the hazard. SE models each control as its own record — forward-only lifecycle (Planned → InPlace → Verified → Removed), audit stamp on verification, colour-coded hierarchy-ordering badge. A failed verification is on a specific control record, not buried in the hazard's edit history. The hazard's Resolved transition gating on Verified controls is structural, not a manager's read of a checkbox.
Industry packs are additive, not exclusive
A logistics-and-manufacturing operator (a contract manufacturer running their own DC network) selects both packs and sees the union — 22 universal + 10 Manufacturing + 6 Logistics — composed additively without conflict. Pack-source tagging is preserved on every record (per-pack visual badges in the admin UI are a future polish). This is the inverse of "pick one industry vertical at provisioning."
AI candidates carry citations
SDS-driven candidate hazard extraction returns each candidate with a grounded flag and a citation back to the source PDF page + paragraph. Approve preserves the citation on the hazard record permanently. When an inspector asks "how did you know this material was a Class 3 flammable?" the answer is "GHS Section 2 of the supplier's SDS, page 1, paragraph 3" — not "the safety officer's professional judgment." That's the evidentiary floor for arbitration and regulatory dispute.
Hazards and incidents share one connected record
An incident's investigation can surface a hazard at the same location with the originating-incident backlink preserved. A future incident at that location auto-suggests the realized-from-hazard linkage. Same corrective-action engine, same root-cause investigation, same task workflow, same audit trail. Not "incident-to-hazard integration" — a single model with two front-doors.
Adjacencies
What hazards connects to.
Incidents
Realized-from-hazard backlink on incidents; originating-incident backlink on hazards born from investigations; RCA spans both subject types.
Explore →
Corrective actions
Same corrective-action engine as incidents; the close-gate prevents a Resolved transition while open tasks remain; discussion threads + evidence attachments on every task.
SDS + chemical safety
SDS uploads trigger candidate hazard extraction with grounded citations; chemical hazards link to the SDS record for ongoing reference.
Explore →
Observations
Observation pipeline is separate (template-driven dual-track compliant / non-compliant); adjacency exists via the Corrective Action engine cross-linking.
Explore →
IndustryPacks
Tenant-selected packs seed pack-specific hazard categories + RCA templates + incident categories; multi-pack composition is additive.
Map view
An interactive site map plots every open hazard per site; resolved hazards drop from the pin layer automatically.
On the roadmap
What's next for hazards & risk.
Everything above is shipped and ready to demo. These are the focused next additions for this area, sequenced by customer demand.
Continue exploring
More on the SE platform.
Five live feature spokes + two roadmap pages + the Workers' Comp claims roadmap. Jump anywhere.
See the hazards register in the running product.
A 30-minute walk-through against your actual hazard shape — your industry vertical(s), your site count, your current register tooling. We'll show the matrix + the hierarchy of controls + RCA-on-a-hazard, and how incidents, corrective actions, and SDSes share one connected record and the same audit trail.