# SoundSafe.ai — full site content > SoundSafe.ai builds correlation engines for physical infrastructure: software that takes > the data a facility already produces, in the incompatible formats it already lives in, and > makes it answer the questions the people running that facility actually ask. Two products > ship on one engine — OmniusGrid for plant and logistics data, and the Security Layer for > camera and sensor feeds. > > This file is the complete text of soundsafe.ai, generated from the site itself so it cannot > drift from what the page says. Generated 2026-09-12. --- ## Home (/) Omnius Grid by ### The correlation engine for Industry 4.0. It takes the files a factory already has — ERP orders, invoices, shipments, quality records — and the machine data it already streams, and makes them answer questions together. In plain language, with a score, the reasoning shown and a one-click action. Sales just called. Big rush order. Can we take it? Book a walkthrough See how it answers Documents only to start. No machine access, no network change.Galena & Chicago, Illinois · Dubuque, Iowa Omnius Grid · Correlation AI live XLSX material-forecast_q3.xlsx PDF carrier-agreement_2026.pdf ERP SAP · work orders LIVE Lines 1–3 stream What happens if we raise Line 3 output by 20%? Working through the analysis Reasoning — ingested files × live machine data GO · 91 Line 3 absorbs +20% without a new bottleneck — materials hold through Q3, freight is already contracted, and Lines 1–2 cover the changeover dip. Green-light it. Green-light +20% on Line 3 Kanban task → Purchasing ✓ dispatched Ask anything about this plant… ➤ Drop in the files you already have. Ask in plain language. Replay One question, three places it lives #### Each source is perfectly legible on its own. Together, they've never met. Purchasing material-forecast_q3.xlsx Steel and bearing replenishment holds a +20% draw through Q3. Logistics carrier-agreement_2026.pdf Two extra outbound loads per week, already covered at contract rate. Machines · live Lines 1–3 stream Line 3 at 63% load — Lines 1–2 have room for the changeover dip. ###### The answer: yes — and here's the headroom Three sources resolved into one capacity picture. Line 3 — capacity check live ● LIVE Line 3 63% load +20% headroom Materials hold through Q3 · freight already contracted · Lines 1–2 cover the changeover dip Capacity found, not bought. The answer was already inside the plant. What it actually looks like #### No abstract dashboards. These are the panes you'd work in. ###### See what's wrong, right now Alarms arrive tied to the machine that raised them — not as a code you look up. Active Alarms live critical Die protection tripped on misfeed Stamping Press #1 · DIE_PROTECTION_TRIP high Wire feed stalled — cell held Welding Cell #1 · WIRE_FEED_FAULT medium Discharge temperature above threshold Air Compressor #1 · DISCHARGE_TEMP_HIGH low Spindle bearing temperature elevated CNC Mill #1 · SPINDLE_WARN ###### Fix it without typing a ticket The work order writes itself from the answer, then goes to a named person. Operations Board In Progress 3 WIP Limit 3 / 5 Corrective Maint. (1) ###### Replace spindle bearing kit (WO-4482) Corrective Overdue Asset SAP work order WO-4482 — bearing kit BOM-2231, Line 3 CNC spindle Progress 35% DA Dev Admin ◷ 1h 30m Jul 10 💬 4 Assign to DA Dev Admin Operations ✓ RO R. Okafor Maintenance ✓ SL S. Lindqvist Quality ✓ ✓ WO-4482 assigned to R. Okafor — Maintenance One approval — the floor and Purchasing move together. Replay ###### Know what to do about it Every answer arrives with the next steps attached, waiting on one approval. Recommended Actions Actionable insight — one click ✓ Release the +20% schedule to Line 3 from Monday ✓ Raise steel replenishment to the Q3 forecast draw ✓ Add two outbound loads per week at contract rate ✓ Pull SAP bearing kit BOM-2231 for WO-4482 Causal confidence 0.0 — association shown, causation not claimed. Supervisor approval required before anything dispatches. ###### Catch the cost while you can still stop it Detention is flagged in the yard, not discovered on next month's invoice. Yard Management live TRL-2291 · Dock 4 arrived 06:12 · free time expired 09:12 $325 TRL-1180 · Dock 7 arrived 07:40 · free time expires in 22m $200 TRL-3007 · Dock 2 released 08:55 · within free time $0 Trailers at risk ⚠ 2 Detention exposure $ $525 ##### Dashboards tell you what happened. An operator needs to know what to do next. So every answer carries a score, the rows it was built from, and an action somebody can approve — not a chart to interpret at the Monday meeting. How OmniusGrid works Two products, two budget lines #### One engine discipline, sold to two different people in your building. They're better together and neither one needs the other. Buy either on its own. Flagship Omnius Grid Bought by operations Answers: can this site take the order, and what breaks first if it does - Correlates spreadsheets, PDFs and ERP records into one picture of the plant. - Eight live ERP integrations, SAP through QuickBooks. Enterprise data leads; machines follow. - 1,349 equipment families across 513 manufacturers, every one integration-graded. - Answers carry source, table and row lineage. A supervisor approves before the floor moves. How OmniusGrid works Separate product ##### The Security Layer Bought by security & EHS Answers: what just happened, and across how many of our sites - Runs on the cameras and VMS you already own — Genetec or Milestone — instead of replacing them. - PPE compliance, safe human-machine interaction and restricted-zone monitoring first. - Resolves every feed at a site into one incident, then correlates across the whole estate. - Hash-chained audit trail, built for an insurance claim or a records request. How the Security Layer works Why you can act on the answer #### Evidence, not alerts. ###### It shows its work Source, table and row lineage on every correlation — the exact join key, the records that matched, and the records that didn't. ###### It refuses to guess Ask it for a cause and it returns a causal confidence of 0.0 with an explicit guardrail. That's a property of the output, not fine print. ###### A human stays in the loop Operational change requires supervisor approval, and action assessment fails closed. When it isn't sure, nothing moves. ###### The record is tamper-evident A hash-chained audit trail, chained per organisation, so what the system concluded and when survives outside scrutiny. Frictionless integration #### Nothing gets ripped out. Nothing new goes on the line. The hardest part of industrial software is usually the installation. We removed it from the front of the process. OmniusGrid starts with the files your plant already produces — so the first answer arrives before anyone touches a machine, opens a port, or schedules downtime. ###### No machine access to begin Documents and ERP records only. No network change, no PLC work, no downtime window, no IT review to clear first. ###### Your ERP, not ours Eight live integrations, SAP through QuickBooks. It reads what you already run rather than asking you to move to a new system. ###### Nothing to standardise first A press from the nineties and a cell installed last quarter arrive on the same collector base class. You don't have to make the floor uniform before you get an answer. ###### It runs above your VMS Genetec or Milestone stay exactly where they are. Correlating across the camera estate is the job — displacing it isn't. 1,349 equipment families, every one integration-graded 513 manufacturers across the catalogue 23 protocol collectors, plus eight live ERP integrations Already integration-graded, among others: Siemens Fanuc Rockwell / Allen-Bradley Haas Mitsubishi DMG MORI ABB KUKA Omron Schneider Electric Beckhoff Mazak Trumpf SICK Yaskawa Motoman Okuma Keyence Bosch Rexroth Atlas Copco Heidenhain Makino Engel Arburg + 489 more Book a walkthrough See how it integrates Why Midwest matters #### Being nearby is a service commitment, not a biographical detail. 30.5% of America's mid-market manufacturing sites sit in our eight states, and every one of them is inside a day's drive — which is what keeps site surveys and calibration economical rather than a line item. We expand outward from that core, not by planting a flag on a coast and hoping the service radius stretches. Where we are, and where this goes The eight states we serve today form one contiguous block, and every site in it is inside a day’s drive. Eight states today — 11,485 mid-market sites, all inside a day's drive The adjacent corridor — next, while proximity still holds Coast to coast — once density supports standing service, not before Mid-market manufacturing sites, by state Establishments at enterprises of 50–499 employees, NAICS 31–33 Ohio 2,258 Michigan 1,978 Illinois 1,836 Wisconsin 1,694 Indiana 1,326 Minnesota 991 Missouri 811 Iowa 591 Eight states 11,485 Source: US Census Bureau, Statistics of U.S. Businesses, 2022, released 10 April 2025. Establishment counts rather than firm counts, because we sell per site. The final row is the eight-state total, drawn to its own full width. Start here #### Bring one question your floor can't answer today. Not a requirements document and not a procurement process — one real question, the kind that currently takes a meeting or a week. Thirty minutes is usually enough to show you whether the answer was already sitting in your own files. Book a walkthrough See how OmniusGrid works --- ## OmniusGrid (/omniusgrid) Omnius Grid by ⚡ The correlation engine for Industry 4.0 ### Unleash the power of data correlation Spreadsheets, PDFs, photos, audio, video and live sensor feeds — one platform, one answer. Ask in plain language and get a risk-scored answer with its reasoning shown and the actions attached. Just ask. It correlates. Book a walkthrough See the Security Layer ###### "What happens if we raise Line 3 output by 20%?" The canonical question, answered end to end. Correlation AI live Reasoning — ingested files × live machine data GO · 91 Line 3 absorbs +20% without a new bottleneck — materials hold through Q3, freight is already contracted, and Lines 1–2 cover the changeover dip. Green-light it. Actionable insight — one click Green-light +20% on Line 3 Kanban task → Purchasing ✓ dispatched Causal confidence 0.0 — association shown, causation not claimed Industry 4.0 #### The fourth industrial revolution runs on data. Enterprise data — ERP orders, invoices, shipments, quality records — holds half the story. Smart machines and smart factories produce the other half. The gains come when the two meet: transparency, flexibility, better decisions. But that data lives scattered — ERP, MES, spreadsheets, cameras, sensors. You cannot make smart decisions with scattered data. Omnius Grid is the solution. Every format — Excel, PDFs, photos, audio, video, live sensors — one context Shop floor → logistics → financials → customer, correlated end to end Ask in plain language; get risk-scored answers with recommended actions On arrival #### Every format becomes correlated data the moment it lands. Excel tabs, PDFs, a whiteboard photo, live feeds — one session, one question. Nothing to wrangle into shape first, and no schema to agree on before you can ask. XLSX Spreadsheets PDF Reports & agreements JPG Whiteboard photos WAV Acoustic monitors RTSP Camera feeds LIVE Sensor telemetry ERP Orders & invoices Every ERP, synced live SAP Oracle NetSuite Dynamics Odoo Infor Epicor QuickBooks Shop floor → logistics → financials → customer, correlated end to end. Patterns you didn't know existed #### Six sources. One chain of events. Seconds, not a shutdown. This is a real investigation surface: the question asked in plain language, and the answer that comes back correlated across every source the plant already had. Line 3 Spindle Degradation — Cross-Source Investigation 6 sources Why did Line 3's scrap rate jump 3.4% last week? I've added the production log, compressor report and whiteboard photo. Critical Risk: 78.0 / 100 telemetry maintenance erp quality logistics finance ###### What happened on Line 3 Scrap rate rose 3.4% between Jul 1–7. Correlating all six sources points to a single chain of events: 1 Logistics — inbound shipment SHP-2214 (spindle bearing stock) arrived 2 days late. Dock Camera — Door 3 motion timestamps confirm material staged after 14:00 on Jul 2. 2 Shop floor — the production log Downtime tab aligns with the QC Defects tab: asset-8 , vibration 7.8 mm/s against a 4.5 threshold, and the Compressor Room acoustic monitor shows a +6 dB rise. The whiteboard note "spindle whine after 13:00" matches the vibration onset. 3 Maintenance — Compressor_Service_Report.pdf pages 4 and 7 flag progressive bearing wear; SAP work order WO-4482 is still open, correlated 0.82 with the vibration trend. 4 Financials and customer — estimated cost impact $18,400 in scrap so far. Invoice INV-55810 risks a rework charge and customer order SO-88231 slips 2 days unless the PM is pulled forward. Every line above carries its source file, table and row · causal confidence 0.0 — association shown, causation not claimed Replay No dashboards to build. No data to wrangle. Just ask — and get the correlated answer. The surfaces you would work in #### The same correlated data, wherever the work happens. An asset on the floor, the yard outside it and the freight between sites are usually three systems that disagree. Here they are one. Every pane below is OmniusGrid itself, running on the demo dataset it ships with. ###### The asset, as the people who run it see it. A stamping press, its live alarms and the OEE the shift is judged on — one screen, not four systems. Assets — Stamping Press #1 machinery · SCH-MC250-01 ### Stamping Press #1 Schuler MC-250 • SCH-MC250-01machinery #### Active alarms(4) Wire feed stalled — cell held high • WIRE_FEED_FAULT • 09/11/2026 09:39:31 PM Acknowledge Discharge temperature above threshold medium • DISCHARGE_TEMP_HIGH • 09/11/2026 08:39:31 PM Acknowledged Spindle bearing temperature elevated low • SPINDLE_WARN • 09/11/2026 07:39:31 PM Acknowledge Die protection tripped on misfeed critical • DIE_PROTECTION_TRIP • 09/10/2026 09:39:31 PM Acknowledged #### OEE (last 24 hours) Availability 91.0% Uptime vs planned time Performance 85.0% Speed vs ideal cycle time Quality 99.0% Good units vs total (1228/1240) OEE 78.0% Availability × Performance × Quality State breakdown · Last 24 hours Execute 8h 0m Idle 27m Held 15m Stopped 3m ###### The numbers underneath it, and the controls beside them. Tonnage, strokes, die temperature and shut height as the press reports them — and the commands that act on what they say. Assets — telemetry & command PROFINET · state Execute #### Latest Telemetry press tonnage 340t 10:39:35 PM strokes per_min 45SPM 10:39:35 PM die temperature 61.4°C 10:39:35 PM shut height 268.4mm 10:39:35 PM counterbalance pressure 4.9bar 10:39:35 PM ##### Command Control Stamping Press #1 State: Execute Emergency Stop STOP NOW Immediately halts all operations. Use with caution. Pause Job Pause current printing/processing job Resume Job Resume paused job Set Speed Adjust print/processing speed Set Temperature Adjust a heated-zone setpoint (barrel, die or oven) Speed % Range: 10 - 200 Send Command Cancel ###### The yard, counted rather than estimated. Trailers, dwell time and the detention it is costing right now — with the dock doors they are waiting on. Yard Management System (YMS) 4 trailers · $525 detention Total Trailers 4 In Yard 2 Docked 1 In Transit 1 Detention Risk 2 Detention Cost $525 Available Doors 1 Today's Appts 2 ##### Detention Alerts (1) GHI-3456 Swift Transportation • ZONE-A-05 $450 4h 0m excess Trailer Status Carrier Location Duration Detention Contents TR-2024-001 ABC-1234 docked Swift Transportation DOCK-A1 2h 0m $0 SL-998877 TR-2024-002 XYZ-5678 yard Schneider National ZONE-B-12 4h 0m $75 SL-998878 TR-2024-003 DEF-9012 in transit JB Hunt - 1h 0m $0 - TR-2024-004 GHI-3456 yard Swift Transportation ZONE-A-05 6h 0m $450 SL-998879 ##### DOCK-A1 — occupied Occupied by: ABC-1234 ##### DOCK-A2 — available ##### DOCK-B1 — reserved ##### DOCK-B2 — maintenance ###### And the freight moving between them. Shipments, carriers and hours-of-service, with the telematics labelled for what they are. Transportation Management System (TMS) 3 shipments · 1 HOS violation Total Shipments 3 In Transit 2 Delivered 0 Planned 1 Active Drivers 2 HOS Violations 1 Carriers 3 CT-PAT Cert 2 ##### Fleet Status (simulated) Simulated telematics. Not measured from a device and not valid for DOT/ELD compliance reporting. Devices 3 Active 2 Drivers 4 On Duty 2 Distance Today 1250 mi Fuel Efficiency 7.4 mpg Shipment Status Carrier Route ETA Freight SHP-2024-0001 PO-78234 in transit Swift Transportation Chicago → Los Angeles 9/12/2026 $2850.00 SHP-2024-0002 PO-78235 in transit Schneider National Denver → Dallas 9/12/2026 $3200.00 SHP-2024-0003 PO-78236 planned JB Hunt Houston → Atlanta 9/14/2026 $4500.00 One record, read four ways. The press, the trailer waiting on its dock and the shipment it is going out on are the same data — which is why the detention clock and the production schedule can finally be compared. The three things it runs #### Optimized operations. Actionable insights. Maximum efficiency. ###### Optimized Operations See the true load, health and headroom of every line — and grow on evidence, not intuition. ● LIVE Line 3 63% load +20% headroom 63% load streaming → +20% green-lit. Capacity found, not bought. ###### Actionable Insights Every answer arrives scored, with recommended actions you assign to a department in one click. GO · 91 Approve +20% on Line 3 Kanban task → Purchasing ✓ dispatched One approval — Purchasing and the floor move together. ###### Maximum Efficiency Freight, yard and delivery capacity in one picture — commit to more volume knowing OTIF holds. TRL-2291 · detention $325 TRL-1180 · at risk $200 TRL-3007 · released $0 Freight and yard say yes before you commit the volume. How an account starts and grows #### Documents first. Machines when you're ready. Most industrial AI wants sensor access on day one, which means a network conversation, an IT review and a quarter of delay. We invert it. Your enterprise data leads — it's already complete, already yours, and it already contains the answer. 1 ###### Land — documents only Spreadsheets, PDFs and ERP records correlated into one picture of the plant. No machine access and no network change, and eight live ERP integrations span SAP through QuickBooks. 2 ###### Expand — the full data surface Machine telemetry joins the picture, then camera feeds. Same engine, same contract, same vendor — no second integration project and nothing to re-procure. 3 ###### Switch on — the Security Layer PPE compliance, safe human-machine interaction and restricted-zone monitoring on cameras that are already installed. A different budget line and usually a different buyer, but nothing new to deploy. What a first engagement looks like #### Prove it on your own history before it ever touches production. You don't have to take our word for any of this, and you don't have to commit a plant to find out. 1 ###### A 30-minute walkthrough Bring one real question from your floor — the one that currently takes a meeting to answer. We show you what the engine does with it. 2 ###### Five weeks on your own records Working only from your historical data, the model has to predict at least three known past production bottlenecks. You already know the answers, so you can score us honestly. 3 ###### A 90-day pilot on one site Scoped before it starts, documents only to begin. Machines and cameras join later, on your schedule. Start with a walkthrough --- ## Security Layer (/security) Unified Threat Engine ### The cameras are already there. Nothing reads them together. Most sites have more coverage than they can use. Every feed raises its own alarms into its own queue, so a pattern repeating across four buildings reads as four unrelated tickets nobody connected. What follows is one night at a single site, from first detection to a sealed case. Book a walkthrough See OmniusGrid Sold standalone. No OmniusGrid deployment required. ###### Live Threat Assessment The score the whole estate is read by, on the engine's own 0–1 scale. UTS Engine — Live Threat Assessment real UI Live Threat Assessment MINIMAL Overall Threat Score 0.23 System Connection Connected Last Update 11:02:51 PM The product's own markup and stylesheet — not a picture of it. 02:00 Detection What the site is watching for #### Thirteen detection layers, switched on per site. Before anything happens, the site decides what it looks for. Licence-plate recognition and vehicle make/model through to aggression, weapons, drone, crowd, intrusion, fire and smoke — each toggled for the site it runs at. Main Feed — Detection Layers 13 active ##### Detection Layers 13 Active ###### LPR License Plate Recognition ###### VMMR Vehicle Make Model ###### Aggression Violent behavior detection ###### Weapons Weapon identification ###### Drone Unauthorized drone tracking ###### Crowd Crowd density analysis ###### Person Person tracking ###### Vehicle Vehicle tracking ###### Loitering Loitering detection ###### Suspicious Suspicious activity ###### Intrusion Perimeter breach ###### Fire Fire/Flame detection ###### Smoke Smoke detection Every layer feeds one clock. Whatever any of these catch lands on the same timeline as everything else, rather than in thirteen separate queues. 02:14 → 07:00 One clock The night, as it happened #### Four lanes, one clock. Audio, video and sensors on a single axis, and a fourth lane carrying what the engine fused out of them. The scrubber runs the night back: unauthorised entry at the substation yard at 03:38 and Transformer T2's partial-discharge rise at 03:49 sit eleven minutes apart on the same line, which is what makes them one story instead of two tickets. Main Feed — Event Timeline · Site A live 02:00 03:00 04:00 05:00 06:00 07:00 Audio Acoustic Turbine 3 bearing whine Acoustic Feedwater pump cavitation Acoustic Conveyor C-7 idler squeal Video Video PPE breach — Line 3 guarding Video Unauthorised entry — substation yard Video Furnace 1 flame instability Video Tanker at loading gantry Video Patrol pass — logged Sensor Sensor Switchgear arc-flash precursor Sensor Transformer T2 partial discharge Sensor CNC Mill #1 spindle 7.8 mm/s Sensor Cooling tower fan imbalance Sensor Compressor discharge overtemp Unified Fused Substation intrusion + PD rise Fused Bearing wear — asset-8 Score Confidence Source: Time: minimal low · 0.3 medium · 0.6 high · 0.8 critical · 0.9 15 events · 3 modalities · fused into one Fifteen events, three modalities, one fused lane. Click any blob for its confidence, model, latency and impact score. A single-site detection vendor shows you three lists and leaves the correlation to whoever is on shift. 03:41 Action What the site was allowed to do #### Call-of-Action and Rules-of-Engagement, as a working board. The column a card lands in is derived from the UTS score, not chosen by hand. Each carries its phases, the person responsible for each action, and the rule of engagement that governs it — including the one action the rules refused. COA / ROE — Call-of-Action board column derived from UTS score Suggested 2 Completed 2 Rejected 1 Suggested 2 Thermographic scan — Transformer T2 Partial-discharge trend correlated with acoustic rise. low preset › 1 phases · 0 complete Scan bushings and radiators · Reliability tech ROE · No de-energisation without grid consent Re-run PD trend against 90-day baseline Confirm whether the rise is drift or a step change. low hybrid › 1 phases · 0 complete Pull historian trend · Condition monitoring ROE · Baseline exports retained 30 days unless attached to a case Pending 2 Isolate switchgear bay 4 Arc-flash precursor detected; bay carries the substation yard feed. high preset human sign-off › 3 phases · 1 complete Confirm load transfer path · Grid controller Rack out bay 4 breaker · HV authorised person Post danger notices · Safety officer ROE · Switching above 11 kV requires two authorised persons Pull loading-bay acoustic 03:15–03:25 Forced-fixture signature, +6 dB over ambient. medium hybrid › 2 phases · 0 complete Export mic-01 segment · Audio analyst Run tamper check · Audio analyst ROE · Audio exports must be hash-chained before review In Progress 2 Dispatch technician to substation yard Unauthorised entry validated across video and access control. critical preset human sign-off › 3 phases · 1 complete Dispatch nearest patrol · Guard 2 Secure yard gate · Guard 2 Log incident to case file · Night supervisor ROE · Observe and report only — escalate above UTS 0.90 Progress 60% Derate CNC Mill #1 spindle Vibration 7.8 mm/s against a 4.5 threshold. high hybrid human sign-off › 3 phases · 1 complete Notify Line 3 supervisor · Ops lead Apply 15% RPM derate · Controls engineer Schedule bearing replacement · Planner ROE · Derate requires ops-lead sign-off before shift change Progress 35% Completed 2 Lock substation perimeter gate Executed 03:41, confirmed by the night supervisor. medium preset › 2 phases · 2 complete Engage maglock · Guard 1 Confirm lock state in ACS · Security desk ROE · Perimeter gates fail open on fire alarm Progress 100% Notify shift engineer Acknowledged at 03:39. low preset › 2 phases · 2 complete Send page · Automated Record acknowledgement · Automated ROE · Critical events page the on-duty engineer within 60 s Progress 100% Rejected 1 Estate-wide load shed Declined — outside the rules of engagement for this tier. critical hybrid human sign-off › 2 phases · 2 complete Assess estate-wide risk · Grid manager Decision: decline load shed · Grid manager ROE · Requires UTS above 0.95 and executive sign-off Isolate switchgear bay 4 Detection → Suggested high PRESET Details Actionables (3) ROE (2) Comments (2) Attachments (1) Dependencies (0) History (4) ###### Description Arc-flash precursor detected on switchgear bay 4; the bay carries the substation yard feed. Correlated with Transformer T2 partial-discharge rise at 03:49. Priority high Status Pending Assigned Unassigned J. Mensah · HV authorised person R. Okafor · Grid controller Safety officer on shift Due Before shift change Template switchgear-isolation Source Detection → Suggested Phase 1 · Assess 1 of 1 Confirm load transfer path Grid controller Phase 2 · Isolate 0 of 1 Rack out bay 4 breaker HV authorised person Phase 3 · Secure 0 of 1 Post danger notices Safety officer ###### Rules of Engagement These constraints define the boundaries and protocols for executing this COA. Modifying ROE requires justification and is logged for compliance. AUTHORIZATION ###### Two authorised persons Switching above 11 kV requires two authorised persons present at the bay. Value: ≥ 2 HV authorised ROE-114 PROHIBITION ###### No estate-wide load shed Estate-wide load shed is prohibited below UTS 0.95 without executive sign-off. Value: UTS ≥ 0.95 ROE-118 THRESHOLD ###### Escalate above UTS 0.90 Observe and report only; escalate to the grid controller above UTS 0.90. Value: UTS 0.90 Modified: 2026-09-11 03:44 — "raised from 0.85 after the 03:27 precursor" ROE-121 Estate-wide load shed was declined. Not forgotten, not quietly executed — refused on the record, because it needs UTS above 0.95 and executive sign-off. Open a card to see its actionables and the rules that govern them. 06:12 The morning after The part nobody demos #### An incident becomes a case, one step at a time. The other half of the job is the case record. The Case Constructor builds it in the open — sector, category, the pipeline that category implies, the case details, the tools it will run with, and the evidence attached to it. Every pane below is that screen, isolated. ###### Every case starts with a sector. Ten of them, each opening its own categories and investigation protocols. An operator hurt at a press is Industrial — manufacturing, chemical plants, refineries and mining. Case Constructor — Select Sector step 0 · sector 0 ##### Select Sector Law Enforcement 4 categories Military 8 categories Critical Infrastructure 8 categories Healthcare 4 categories Financial 4 categories Industrial 4 categories Pharmaceutical 4 categories Educational Facilities 4 categories Aviation Security 4 categories Maritime Port Security 4 categories Selected: Industrial Manufacturing, chemical plants, refineries, and mining operations 1 ##### Select Case Category ##### Select Case Category 4 Categories 1 Selected ###### Workplace Safety Incidents Selected OSHA violations, workplace accidents, injuries, and safety protocol failures. ###### Case Typologies: 2 Available WSI-01: Workplace Safety Incident Investigation 4 Indicators WSI-02: OSHA Violation Investigation 4 Indicators ###### Case Types: 8 Types OSHA Violations Workplace Accidents Injury Incidents +5 more ###### Cross-links: 4 Links OSHA Workers Compensation +2 more ###### Equipment Failures & Malfunctions Equipment breakdowns, production line failures, maintenance issues, and system malfunctions. ###### Case Typologies: 3 Available EF-01: Equipment Breakdown Investigation 4 Indicators EF-02: Production Line Failure Investigation 4 Indicators +1 more typologies ###### Case Types: 8 Types Equipment Breakdowns Production Line Failures Maintenance Failures +5 more ###### Cross-links: 4 Links Maintenance Teams Equipment Manufacturers +2 more ###### Environmental Incidents Chemical spills, emissions violations, environmental compliance, and hazardous material incidents. ###### Case Typologies: 3 Available EI-01: Chemical Spill Investigation 4 Indicators EI-02: Air Emissions Violation Investigation 4 Indicators +1 more typologies ###### Case Types: 8 Types Chemical Spills Emissions Violations Waste Disposal Issues +5 more ###### Cross-links: 4 Links EPA Environmental Services +2 more ###### Quality Control & Product Issues Defective products, recalls, quality failures, and manufacturing defects affecting product safety and compliance. ###### Case Typologies: 3 Available QCI-01: Product Recall Investigation 4 Indicators QCI-02: Manufacturing Defect Investigation 4 Indicators +1 more typologies ###### Case Types: 8 Types Defective Products Product Recalls Quality Control Failures +5 more ###### Cross-links: 4 Links Quality Assurance Customer Service +2 more ##### Subsystem Details 2 Typologies Available Pipeline Active ###### Case Typologies 0 Applied 0 Running 2 Ready ###### WSI-01Workplace Safety Incident Investigation Comprehensive investigation of workplace safety incidents including OSHA violations, accidents, injuries, and safety protocol failures. ▶ Ready 4 indicators 4 methods 4 evidence Apply Comprehensive investigation of workplace safety incidents including OSHA violations, accidents, injuries, and safety protocol failures. ###### WSI-02OSHA Violation Investigation Investigation of OSHA regulatory violations including citation responses, compliance audits, and safety standard enforcement. ▶ Ready 4 indicators 4 methods 4 evidence Apply Investigation of OSHA regulatory violations including citation responses, compliance audits, and safety standard enforcement. ###### Investigation Pipeline Configuration Configure the analysis pipeline for this investigation 0 Enabled 0 Configured Pipeline Active ###### Data Sources Select data sources for analysis ○ Pending • OSHA incident reports • Employee injury records • Safety inspection logs • Equipment maintenance records +3 more sources ###### Analysis Methods Configure analysis techniques ○ Pending • Root cause analysis • Safety protocol review • Equipment failure analysis • Human factors assessment +2 more methods ###### Evidence Containers Configure evidence handling ○ Pending • OSHA incident investigation reports • Safety violation documentation • Injury reports and medical records • Equipment inspection reports +2 more containers 2 ##### Case Details Provide the basic information for your investigation case. ##### Case Details Case Number Priority Low Medium High Case Title Description Assigned To 3 ##### Investigation Tools Select the tools and analysis methods you want to include in this case. ##### Investigation Tools Vehicle Search Search for vehicles by license plate, make, or color Audio Analysis FLAGSHIP Audio Intelligence Engine - G-Host and MAIA integration for comprehensive audio forensics, deepfake detection, and investigative analysis V-Print (TM) Transcription Load audio files and transcribe with V-Print (TM) AI Behavior Analysis Detect suspicious movements and behavioral patterns Cross-Camera Analysis Track movements across multiple camera feeds Sensor Correlation Correlate sensor data with audio/video events Timeline Reconstruction Build comprehensive event timelines 4 ##### Evidence & Files Upload evidence files, documents, or media related to this case. ##### Evidence Container Add Evidence ###### Evidence Items 0 items No evidence items added yet Click "Add Evidence" to get started ###### Evidence Items 1 items Die-area CCTV — 14:07:22, light curtain muted Video Evidence pending ##### Add Evidence Evidence Type Select evidence type Video Evidence Audio Evidence Document Evidence Sensor Data Image Evidence Witness Statement Evidence Name Cancel Add Evidence ##### Case Summary Sector: Industrial Category: Workplace Safety Incidents Case Number: ACC-2026-114-007 Title: Press line 4 — operator injury at an unguarded intervention Assigned To: R. Okafor · EHS lead Priority: medium Tools Selected: 4 Cancel Create Case Nothing here is a mock-up. Each pane is the Case Constructor's own interface, captured from the dashboard with the wizard walked through end to end. Underneath Configuration How the engine is put together #### The engine is a graph you can open, not a sensitivity slider. Collectors, processors and outputs are nodes on a canvas. Each reports its own throughput and latency, the processor fuses them, and the output node carries the score every other surface reads. Add a modality and you add a node. UTS Engine — Node Configuration 5 nodes · 4 connections #### Node Library Audio Source Captures and processes audio intelligence Video Source Captures and processes video intelligence Sensor Source Captures and processes sensor data Processor Processes and analyzes intelligence data UTS Output Unified Threat Score output PTZ Camera Pan-Tilt-Zoom camera with intelligent control Stitching Processor Real-time panorama stitching from multiple feeds Control Module Intelligent control for PTZ cameras and other devices 5 nodes • 4 connections Undo Redo Save Load Execute Clear All Configuration: Apply Zoom: 1:1 100% Select Awaiting telemetry Awaiting telemetry Awaiting telemetry Awaiting telemetry output UTS Output System: UNKNOWN 0.0% Threat Score MINIMAL Connection: Disconnected Total feeds 0 Active 0 Alert 0 Error 0 Intelligence Scores: System Load: 0% Detections: 0 Avg Confidence: 0.0% Correlations: 0 Last: 12:41:54 AM Output Type API Endpoint Database Alert System Dashboard Notification Service Reporting Engine Visualization Control System Logging Service External Integration Routed Feeds 0 No feeds connected audio Audio Sensor Type Microphone Ultrasound Hydrophone Vibration Mic Directional Mic Ambient Noise Infrasound Acoustic Emission Sonar Seismic Voice Recognition Audio Source 1 Status: STANDBY Throughput 2.1 Mbps Latency 12 ms Last: 12:41:50 AM Net: no data Detect: no data video Video Sensor Type Camera Feed Infrared Night Vision Thermal Motion Detection CCTV Satellite Drone LiDAR Depth Camera 360° Camera Hyperspectral Feed Select feed… Iowa DOT: QCTV18 Iowa DOT: CRTV52 Iowa DOT: DMTV66 WI DOT: CCTV-13 MN DOT: C30001 MN DOT: C30047 Video Source 1 Status: STANDBY Throughput 24 Mbps Latency 38 ms Last: 12:41:53 AM Net: no data Detect: no data sensor Sensor Sensor Type Temperature Light Vibration Door/Window Contact Motion Pressure Humidity Smoke/Gas Water Level Proximity Accelerometer Gyroscope Magnetometer Radiation Air Quality Sensor Source 1 Status: STANDBY Throughput 0.4 Mbps Latency 6 ms Last: 12:41:49 AM Net: no data Detect: no data processor Processor Type Machine Learning Signal Processing Image Recognition Audio Analysis Pattern Detection Anomaly Detection Fusion Algorithm Neural Network Statistical Analysis Frequency Analysis Processor 1 Status: STANDBY Throughput 26 Mbps Latency 44 ms Last: 12:41:52 AM Net: no data Detect: no data Eight source types in the library. Adding audio to a video-only platform is a rebuild; here it is a node. Correlation resolves upward #### Three tiers, each seeing further than the last. The same feeds, read at three heights. Most vendors stop at the first one. Tier 1 ###### Smart detection, per feed PPE compliance, safe human-machine interaction, restricted-zone and trespass monitoring, person and vehicle tracking, licence plate. Then aggression, weapon and crowd-surge detection on the same cameras. 200+ detection types 750+ detection components <100 ms detect to response Digital PTZ tracking, PID control Tier 2 ###### Site correlation Every feed at one site resolved into a single incident rather than a queue of separate alarms — running over the VMS you already have, Genetec or Milestone, rather than replacing it. Runs over Genetec / Milestone 1,000+ concurrent sensors Live video, HLS streaming Multi-protocol IoT intake Tier 3 ###### Estate correlation The whole estate at once, so a pattern repeating across four buildings reads as one campaign rather than four tickets nobody connected. A single-site detection vendor cannot do this. Self-healing event bus Predictive & geospatial analytics Estate-wide threat scoring Auto-scaling, heartbeat failover Compliance SOC 2 · GDPR · HIPAA Architecture Zero-trust Scale 1,000+ sensors, auto-scaling Response Sub-100 ms The same discipline, different stakes #### What carries over from the operations engine. Evidence lineage, the causation guardrail, human approval and the hash-chained audit trail are properties of the engine itself, so this layer inherits all of them. In operations that discipline settles an argument. Here it is what lets a record survive a legal challenge, an insurance claim or a public-records request. ###### It runs on your incumbent VMS Displacing Genetec or Milestone isn't the sale — correlating across them is. Your existing camera investment stays exactly where it is. ###### It feeds operations, if you run both What the cameras and sensors resolve becomes part of the operational picture in OmniusGrid. Same building, same vendor, no second integration. Talk to us about your estate Unified Threat Engine #### Bring us one night your team had to reconstruct by hand. An incident that took someone an afternoon of scrubbing four camera systems to piece together. That's the thing worth showing, and it's the fastest way to see whether correlating across your estate is worth a budget line. Book a walkthrough See OmniusGrid --- ## Company (/company) Company ### Midwest operators, building for Midwest plants. SoundSafe.ai builds correlation engines for physical infrastructure. We aren't another chatbot company — the work is unglamorous and specific: take the data a facility already produces, in all the incompatible formats it already lives in, and make it answer the questions the people running that facility actually ask. Everything we ship comes out of one engine, which is why a second product costs a customer a switch rather than a second deployment. Founded January 2024 · Galena & Chicago, Illinois · Dubuque, Iowa What we build · the industrial line #### One engine. Two products in market today, and the same engine underneath both. Correlation is architectural, not a feature you add later — which is why these two share a spine instead of sharing a logo. Each resolves a different picture for a different buyer. Other product lines run on their own track, and you'll find those further down. Unified Correlation Engine ###### OmniusGrid Resolves a capacity picture Can this site take the order, and what breaks first if it does? Documents and ERP records lead, machine telemetry follows. Bought by operations. See OmniusGrid Unified Threat Engine ###### The Security Layer Resolves a threat picture What just happened, and across how many of our sites? Cameras and audio lead, records follow. Runs over the VMS already installed, and sells standalone to a different buyer. See the Security Layer Evidence lineage, the causation guardrail, human approval and the hash-chained audit trail are properties of the engine itself — so every product inherits them rather than re-implementing them. Run both and each one sees more: what the cameras resolve becomes part of the operational picture. Who you'd be working with #### Four exceptional leads, and no fluff. ###### Hamad Dada CEO, co-founder Leads development across operational intelligence, combining technical leadership with domain expertise. Former Operational Director at Sears, so the physical supply chains this deploys into are understood from the inside. ###### Johnny Stevenson COO, co-founder 16+ years scaling businesses and industrial operations, and co-founder of Invasive Media. Owns commercialization, partnerships and market expansion. ###### Dr. Justin Sinkovich Chief Strategy Officer Technology strategist in organizational transformation with a doctorate from USC. Co-founder of Epitonic, a Webby Award winner, and Associate Dean of the School of Business and Entrepreneurship at Columbia College Chicago. ###### William R. Gray Chief Financial Officer Certified Public Accountant, MST, and CVA candidate. Series 7 and 66, licensed insurance producer with Regulation A+ experience. Formerly served in a United States Army intelligence unit. The engineering bench #### Six university partnerships across three states. Applied research, interns and doctoral collaboration — how a Midwest company staffs serious engineering at a fraction of coastal cost. University of Chicago Illinois Institute of Technology Columbia College Chicago University of Wisconsin–Madison University of Dubuque Loras College Galena, IL Where the platform is built Chicago, IL Commercial and university partnerships Dubuque, IA Tri-state manufacturing corridor Within a day's drive of 11,485 mid-market manufacturing sites — which is what keeps site surveys and calibration economical rather than a line item. In development #### More is being built — some of it feeds OmniusGrid, some of it runs on its own track. Two different things are in flight and it is worth keeping them apart. Some modules land inside OmniusGrid and make the answers it already gives sharper. Others are separate product lines with their own buyers, their own roadmaps and no dependency on an OmniusGrid deployment at all. None of it is generally available yet, and all of it is further along than a roadmap slide. Separate product lines — their own path Over Peak Vision In development ###### Inspection Intelligence at Scale Computer-vision damage detection across roofing, concrete and solar — correlated into insurance-grade evidence rather than a folder of photographs. Its own product, its own buyers and its own roadmap — sold to inspection and insurance, not to a plant floor. Roofing · Infrastructure · Solar ReadinessQ In development ###### RQ — the Readiness Quotient A 0–100 score for how likely an individual, team or unit is to execute its intended function to standard under current conditions. A separate line entirely: people and readiness rather than machines and throughput, with its own dashboard and its own customers. Scoring engine · RQ Command · grounded fact surface Modules that feed OmniusGrid Freight Hawk Feeds OmniusGrid ###### Freight Intel & Compliance Suite Freight inspection today is manual and error-prone, which leaves compliance gaps and no verifiable record of cargo integrity. FreightHawk reads the lane with video intelligence instead — car types, ID numbers, cargo counts and securement detail down to strap count and condition, captured and analysed in real time. Guesswork becomes an audit-ready record. → Rail first · architected for road, sea and space ShelfLink OCR Feeds OmniusGrid ###### Automated Shelf Label Capture and Extraction Warehouse inventory stalls on manual data entry — someone walking the racks writing labels down or scanning them one at a time. ShelfLinkOCR detects, reads and extracts shelf labels from video instead, then sorts and curates the data for export. A slow analog audit becomes a high-speed digital one. → Pointcloud stitching · inventory records the engine can correlate ###### A module arrives as a switch FreightHawk and ShelfOCR land inside an OmniusGrid account you already have. The engine, the connectors and the audit trail are in place, so there is no new integration project, no second procurement and nothing new on the floor. ###### A separate line stands on its own OverPeak Vision and ReadinessQ are not OmniusGrid features and do not need it deployed. They are their own products with their own buyers — built by the same team, held to the same evidence discipline. Ask for an early look See OmniusGrid --- ## FAQ (/faq) FAQ ### The questions we actually get asked. Written the way we'd answer them on a call, including the ones where the honest answer is "not yet" or "that depends". Getting started #### Before anything is installed. Nothing in the first phase touches a machine, a network or a camera. ###### What do you need from us to start? Files you already produce. A production log, a maintenance report, a carrier agreement, an ERP export — whatever covers the question you want answered. No machine access, no network change, no downtime window, and no IT review to clear first. ###### How long until we see something real? A 30-minute walkthrough on your own question comes first. After that, the five-week test: working only from your historical records, the model has to predict at least three known past production bottlenecks. You already know the answers, so you can score it honestly — and fail us on it. ###### Do we have to replace our ERP, MES or VMS? No. That is close to the opposite of the product. OmniusGrid reads your ERP through eight live integrations — SAP through QuickBooks — and the Security Layer runs above Genetec or Milestone rather than displacing them. Correlating across what you own is the job. ###### Our equipment is old and nothing speaks the same protocol. That is the normal case, not a blocker. Hardware abstraction sits underneath the engine, so a press from the nineties and a cell installed last quarter land on the same collector base class — 1,349 equipment families across 513 manufacturers, each integration-graded, over 23 protocol collectors. Trusting the answer #### Why an operator can act on it. The hard questions are about evidence, not accuracy claims. ###### How do we know the answer isn't made up? Every correlation carries source, table and row lineage — the exact join key, the records that matched and the records that didn't, plus coverage and selectivity metrics. You can trace any sentence in an answer back to the row it came from. ###### Does it tell us what caused a problem? No, and it says so explicitly. Ask for a cause and it returns a causal confidence of 0.0 with a guardrail attached. It will show you that scrap rose within 40 minutes of every compressor excursion; it will not claim the compressor caused it. That distinction is a property of the output rather than a disclaimer. ###### Can it change something on the floor by itself? Not without a person. Operational change requires supervisor approval, and action assessment fails closed — when the engine isn't sure, nothing moves. ###### Will the record hold up in an audit or a claim? It is built for that rather than retrofitted for it. The audit trail is hash-chained per organisation, so what the system concluded and when it concluded it is tamper-evident — which matters more on the security side, where a record may face a legal challenge, an insurance claim or a public-records request. Commercials and company #### The practical ones. Where the honest answer is sometimes "ask us". ###### What does it cost? It depends on the number of sites, which surfaces you switch on and the contract length, so a number quoted here would be wrong for most people reading it. We price a pilot before it starts and we don't send a proposal you need a decoder ring to read. Ask and we'll give you the real figure for your situation. ###### Do we have to buy both products? No. OmniusGrid and the Security Layer are separate products on separate budget lines, usually bought by different people. Either one stands alone. Run both and each sees more — what the cameras resolve becomes part of the operational picture — but neither requires the other. ###### Who owns our data? You do. It stays yours, it is chained per organisation in the audit trail, and correlating it is the service — not the product we sell to someone else. ###### You're a small team. What happens if we depend on this? Fair question, and the reason we don't sell ahead of what we can support. An operator whose floor depends on this needs a standing service level, which takes people who exist before the first incident rather than after it. That is what we're staffing for, and it's why we'd rather start you on the five-week test than a signature. Still unanswered #### Ask us the one that isn't on this page. The awkward questions are the useful ones, and they reach the people building this rather than a form queue. Ask a question Read the notes --- ## Notes (/notes) Notes ### What we're learning, written down as we go. Not a content programme. Notes from building an industrial correlation engine — what broke, what the data actually said, and the decisions we'd defend in a room full of operators. OmniusGrid Coming soon ##### Why we refuse to tell you what caused it Every industrial AI demo claims root cause. Ours returns a causal confidence of 0.0 and shows the association instead. Here's the argument for why that's the more useful answer on a plant floor — and why a vendor who claims causation from observational data is telling you something they cannot know. Draft in progress OmniusGrid Coming soon ##### The order you already had capacity for A walk through one correlation end to end: a rush order, four data sources that never meet, and the seventeen points of headroom nobody could see. The interesting part isn't the answer — it's how long the answer had been sitting there. Draft in progress Security Layer Coming soon ##### Four tickets or one campaign A single-site detection vendor sees four unrelated alarms across four buildings. What it takes — architecturally — to see one pattern instead, and why correlation across an estate can't be added to a per-camera product later. Draft in progress Engineering Coming soon ##### Tests that check our own marketing Published figures go stale the moment the work moves on, and nobody notices. We wrote tests whose only job is to check the numbers in our own documents against the code that produces them — if a published count stops matching what shipped, the build fails. Draft in progress We'd rather publish four things worth reading than forty for a search engine. If there's something you want us to write up, say so — it moves up the list. Suggest a topic --- ## Contact (/contact) Contact ### Tell us what you'd ask it. The most useful first conversation is a real question from your floor — the one that currently takes a meeting to answer. Bring that, and we'll show you what the engine does with it. No deck required. Leave this field empty Your name Company and site location Email The question you'd want answered Send us the question Sent straight to us — no mail app needed. Submissions are handled by Netlify, our host. Or email us directly info@soundsafe.ai Where we are Galena, IllinoisChicago, IllinoisDubuque, Iowa What to expect A 30-minute walkthroughThen five weeks on your own dataThen a 90-day pilot on one site