The headline of this release. Storyboard what an operator has to do by dragging eight plain-language blocks - Look, Listen, Recall, Decide, Act, Speak, Wait, Subtask - onto a timeline, and ErgoACT simulates the task on a real ACT-R cognitive architecture. No modelling jargon is required, though an ACT-R mapping expander and a read-only generated Lisp model are there for those who want them. Monte Carlo runs return mean and spread of completion time, a 5th to 95th percentile range, a completion-time histogram, per-step time share, and which mental channel - vision, memory, hands or speech - is the true bottleneck. Operator profile sliders for skill, fatigue, stress and age genuinely move the timings, and every run is reproducible from its seed.

Steps can be flagged to run alongside the one before them. Work that genuinely can overlap does; work competing for the same mental channel queues, and dual-task performance degrades under stress. The reported "time saved by multitasking" figure refuses to invent savings that the model did not earn. Each Recall step declares what happens when memory fails - guess and continue, consult a job aid at a real recovery cost, skip the dependent steps, or abort - so every run ends Success, Completed with errors, Completed with omissions, or Aborted, yielding a task-level human error exposure rate derived from cognitive first principles to sit beside HEART, THERP and SPAR-H. Pin a baseline and read green and red deltas against it, replay the median run on the storyboard, chain saved models together as subtasks, and import steps straight from a linked HTA. The Word report ships with a dedicated reading guide explaining every figure and what a tight peak, a long tail or twin peaks actually mean.

Expert-judgement estimation of human error probability. Rate a set of comparable tasks against Performance Shaping Factors and SLIM computes a Success Likelihood Index for each, converting it to an error probability by calibrating against reference tasks with known rates. Ships with a starter PSF library with cited definitions, weight normalisation, and support for factors where more is not always better - stress, for example, with an optimal-point marker. The sign-off gate is faithful to the method: you cannot finalise on fewer than two reference tasks, and at exactly two an unmissable warning explains that a two-point fit is perfect by construction and is not evidence of calibration quality. Tasks import from a linked HTA through a selective leaf picker - deliberately with no "select all", because the method depends on choosing a comparable subset. Based on NUREG/CR-3518 (Embrey et al., 1984).

Classify what cognitive errors occurred, or could occur, and why. Both modes ship together: Retrospective for analysing an incident that happened, and Predictive for walking a task and predicting credible errors. A guided decision-flow wizard - the product's first branching question-flow navigator - asks yes or no questions to lead the analyst to the right internal error mode. TRACEr-lite taxonomies are the default, with an Extended toggle adding external error modes plus error detection and correction. Predictive mode presents the linked HTA as a full-width worksheet with parent-goal context, indented steps and an "N of M steps reviewed" progress summary, capturing credible errors as chips with their recovery steps. Grouping sheet-tabs tame very large task trees and survive HTA re-imports. Ships with Word reports for both modes and two worked sample sessions.

Take a defensible position on whether an AI-automation design is acceptable. Declare, for each design stage, a Level of Automation, a Technology Readiness range and a Human Readiness range, then rate ten Key Performance Areas - Safety, Resilience, Security, Human Factors, Accountability, Responsibility, Liability, Efficiency, and societal and environmental sustainability - as Acceptable, Tolerable or Unacceptable on a ten-spoke, three-band radial. Two stages sit side by side with a movement note per area, and critical scenarios and certification objectives carry the reasoning the bands cannot. ErgoCompass is the platform's first tool that consumes rather than produces: it holds only its own ratings and reasoning, and cites its evidence by reference from assessments already saved elsewhere in the project. That evidence is handled honestly rather than optimistically - a cited source that has been deleted or retitled shows as a dangling reference you can re-point or remove, and the rating itself stands, because it was made on the evidence as it was cited. Deliberately, four of the ten areas have no evidence producer among the shipping tools and rely on your own reasoning and external sources; equally deliberately, there is no overall number, because the method defines none and inventing one would be fiction. An unrated area is a finding, not a gap in the tool. Based on Stroeve, Kirwan and Everdij (2026), Safety 12(4) 91, with SESAR, NASA TRL and ANSI/HFES 400-2021 HRL scales.

Alarm rationalisation runs backwards. Alarms get configured, the plant starts running, operators drown in nuisance alarms, and only then does anyone ask what the people responding actually needed to know. ErgoAIR sits upstream of the master alarm database, in the identification and rationalisation stages of the ISA-18.2 lifecycle, and captures the failures an operator must respond to, the indication that tells them it is happening, and the response they take - feeding the control and instrumentation philosophy rather than arguing with it after commissioning. Priority is never typed: it resolves from a severity by time-to-respond matrix you can tune per document. A visible threshold separates what genuinely warrants an alarm from what does not, and everything below it is reclassified as an alert, a prompt, an event-log entry or a suppression rather than deleted, because knowing what to leave out is as valuable as knowing what to instrument. Alarms above the line count against a per-asset budget, named upset scenarios test the load against a first-ten-minutes ceiling drawn from the EEMUA 191 benchmark of roughly one alarm per operator per ten minutes, and shared indications and shared responders are surfaced as single points of failure - one sensor serving four alarms, or one field crew owing three concurrent responses. The consequence tree is computed from the worksheet rather than drawn beside it, so the two cannot drift; it is called a consequence tree and not a fault tree because it performs no Boolean gate logic and will not pretend otherwise. Requirements import from an SCTA study's critical steps, carrying the trace reference, so a stated priority is defensible because it was extracted through a structured process rather than asked cold. Based on EEMUA 191, 4th edition (2024) and ISA-18.2 / IEC 62682.

Measure how mentally loaded an operator is from EEG. ErgoNeuro is built on a genuine, unit-tested signal-processing core: FFT-based spectral analysis producing the five standard bands (Delta, Theta, Alpha, Beta, Gamma), validated workload indices including the Pope et al. (1995) engagement index, theta-alpha workload and theta-beta ratio, and a single 0 to 10 cognitive load index. It works with no hardware at all through a demo stream, records several takes inside one session without overwriting the last, and replays any take with a scrubber and a tap-to-seek peak strip. Live view shows raw channel waveforms, band-power bars or a workload timeline. Word and HTML reports include a takes index, band-power distribution, high-load periods and channel signal quality, with an explicit warning when data quality is too poor for the numbers to be relied upon.

ErgoEvac moves from a locked concept to a working egress simulator. Draw or import a floor plan, place walls, exits, signs, obstacles and spawn points, and simulate. Results are real and reproducible per seed: time to first and last agent through each exit, average and peak flow rates, congestion zones, bottlenecks, unseen signs, and P95 and P99 confidence intervals. Set a target egress time (RSET) and results report a RAG margin against it. Exits can be blocked from the outset or mid-run, with occupants re-routing. Pre-movement time is now distributed log-normally per PD 7974-6 rather than every occupant starting at the same instant, and a playback banner explains the initial stillness that previously looked like a frozen simulation.

Exit signs are modelled to EN 1838, with legibility distance derived from panel height and illumination, a 60 degree viewing cone, and wall or flag mounting that snaps to walls. A coverage overlay maps green catchment against red dead zones with a live coverage percentage, and can be shown under design or under actual dark or smoke conditions. A placement optimiser proposes sign positions to reach around 95 percent coverage as ghost signs you accept individually or all at once, and flags redundant existing signs. Analysis adds Fruin walkway Level of Service bands from peak crowd density, occupant load suggestions from walkable floor area, and a travel-time isochrone overlay showing seconds to the nearest exit. Pin a run as baseline A, change the scenario and re-run to read signed deltas on every headline figure; click any occupant during playback to trace their route, exit choices and how much of their journey was unguided; and run "Find decision points" to mark the junctions where unguided occupants moved away from the nearest usable exit - in other words, exactly where a sign is missing.

All seventeen ErgoTools now walk you through themselves: eighteen tours and 194 steps, ErgoGlare carrying two because it has two editors. A tour dims the window, spotlights the control it is describing, and puts a coach card beside it with Skip, Back and Next; Escape ends it and the arrow keys navigate. Each step sets up the state it describes, and does so in a way that can be re-run, so Back lands you somewhere just as coherent as Next - a tour is not a recording you can fall out of the middle of. Steps anchor to controls by identity and measure them live, so the tour follows the layout rather than a set of remembered coordinates, and a step whose target genuinely is not there drops out rather than spotlighting empty space. Tours are yours to start, replay and abandon: nothing auto-launches, nothing needs suppressing, and nothing is remembered afterwards.

A tour of an empty session has nothing to point at. Eleven tours therefore swap in a canonical worked example for the duration - ErgoAIR's BPS-14 asset, ErgoCompass's Alicante runway-sequencing case, ErgoCWA's Sydney Trains scenario, ErgoEvac's metro platform - and restore your real document the instant the tour ends, however it ends. The example lives in memory only: it is blocked at the point where anything would be written, so it cannot reach your project, your dashboard or your session list. If your session already has content, the tour walks your own data instead. ErgoNeuro tours on its live demo signal, and the six tours whose tools hold state that cannot be swapped safely - ErgoEnsemble, ErgoDesign, both ErgoGlare editors, ErgoLens and ErgoSense - narrate over whatever you have open and point you at the seeded sample sessions instead.

Help is no longer a dropdown hiding beside View. It is a word at the end of the ribbon row that swaps the ribbon in place, as Word's does, leaving what you are working on untouched: two tour buttons and six help buttons - Tool Library and Guides, Keyboard Shortcuts, Check for Updates, Send Feedback, Report a Bug and About ErgoSphere - reading from the same source as the shell menu, so the two cannot drift apart. Fifteen of the seventeen ErgoTools carry it. Alongside See Tool Tour sits Play Tour, which advances on a timer, hands-free, for recording a tutorial without a cursor wandering across the screen. The method tools get the same Help pane without the tour group, which is absent rather than present and disabled - there are no dead buttons and no gap where they would have been. ErgoLens and ErgoSense are the deliberate exception: their tabs navigate between pages, which would end a tour mid-step, so they keep a Help dropdown and start their tours from the File menu.

ErgoSense now renders its own native 3D preview of the live suit, driven directly by the motion data, in place of decoding video frames streamed from an external engine that was punishing lighter laptops. Biomechanics and scoring have moved in-process alongside it, verified in a live suit session, which means an Xsens capture no longer requires the separate ErgoSense CORE application to be running at all. The mannequin itself has been substantially improved: stance feet now plant properly instead of drifting, floating and sliding, trunk bend is distributed across the spine rather than creasing at a single joint, and motion is interpolated for smooth playback.

Moving biomechanics in-process completes the live Xsens data path: RULA and REBA scores now stream valid throughout a capture, verified in a live suit session. A calibration pass went through the dials alongside it - head rotation, trunk flexion and the left leg now track in the correct direction, forearm pronation and supination are re-zeroed so REBA's wrist modifier and RULA's wrist twist rest neutral, and shoulder elevation uses a formula immune to smearing. Trunk extension was verified against a live lean-back and tracks correctly as a negative angle.

ErgoLens now measures the shoulder against the subject's own trunk rather than against gravity or the camera, deriving every axis from the landmarks themselves so the measurement holds however the subject is oriented in frame. A new elevation channel carries the full range from 0 to 180 degrees, which means raised and overhead work is measured across the whole of its range and scored the way an assessor scores it: total humeral elevation is what the eye judges from a photograph, it is what ISO 11226 uses, and it now drives the upper-arm score in RULA and REBA. On a reference photograph of a subject with one arm raised overhead and the other resting at her side, ErgoLens reports 179 degrees and 4.5 degrees and scores those arms RULA 5 and RULA 1 - the same scores an assessor reads from the same image. The trunk is measured through the same body frame, so deep forward flexion is scored across its full range too. The rebuild is proven against synthetic fixtures built at real anthropometric proportions, signed sweeps across the full range of motion, and reference imagery.

The rebuild carries a principle worth stating plainly: a camera that cannot observe a rotation does not report one. Sagittal flexion cannot be seen from a front-on view, so ErgoLens reports zero and tells you why on the dial itself - "not observable from this camera angle" - rather than offering a number the geometry cannot support. Neck rotation is estimated from nose position along the ear line, with a deadband set wide enough that landmark noise cannot trip RULA and REBA's "plus one if twisted" gate. Trunk axial rotation reports zero from a single camera, because a two-dimensional estimate of it would be a guess wearing a number. The per-joint heat strip and the overall score read from the same values, so what the strip shows and what the score says always agree.

ErgoLens gains a full Word report for the first time, with capture snapshots embedded, filed automatically to the Document Registry as compliance evidence.

The AB4D engine introduced in 1.3 now carries the whole of ErgoDesign. The last of the legacy renderers is retired, and the 3D view becomes a place you can genuinely work: click-to-select with move and rotate gizmos, locked items that stay locked, item colour overrides carried through into the scene, and a view that follows the active theme. Real imported 3D models replace the old primitives for monitors, laptops, keyboards, mice and the human manikin, and the manikin now scales correctly against its anthropometry. If the GPU cannot initialise the engine, the application says so in the status bar and degrades gracefully.

The top and side views are now true orthographic projections of the same 3D scene rather than a separate approximation, so a rotated item foreshortens correctly and the three views can no longer disagree with one another or with the saved document. Analytics - the grid, dimension lines, reach zones, ERP markers and the glare heatmap - draw as a transparent overlay above that scene. Report snapshots render from the same engine, so the images in a report finally match what is on screen, with selection boxes and gizmos excluded. The measure tool, previously a non-functional stub, is rebuilt: click and drag to measure, snap to the grid, and read distance in millimetres or metres alongside view-plane components and angle.

Interior obstructions are now real physical participants in the daylight simulation: they occlude and inter-reflect, each carrying its own surface material and reflectance, rather than being cosmetic context. They render in the 3D room and as footprints on the 2D plan behind a single "Show furniture" toggle, though the simulation always includes them.

A workstation laid out in ErgoDesign can be pulled straight into ErgoGlare as an indoor daylighting assessment. The room footprint is sized around the layout, the observer eye point is taken from the ErgoDesign anthropometry, and every piece of equipment becomes an obstruction with a reflectance drawn from a reviewed material table. An "Edit furniture" flyout lists each imported item with a per-item material override and a remove button. Note that the 3D scale handle is retired with the engine change - items are now scaled numerically in the properties panel.

The structural fix practitioners asked for: one task can now carry many use errors. Previously each row was a single task and error pairing, so analysts duplicated task columns by hand. Existing files migrate automatically with no data loss. A critical-task override lets you force a task critical or not critical with a recorded rationale, as IEC 62366-1 expects reasoned identification rather than an automatic severity threshold, and that rationale carries into reports. User Groups and Use Environments are now editable, and a new Known Use Problems card gives post-market data, adverse-event databases and literature the home they previously lacked. The editor is rebuilt as task blocks with indented error rows in place of a seventeen-column flat grid, and reports now auto-generate a draft summative evaluation protocol from the critical tasks - one scenario per critical task, with the errors to challenge, associated harms and acceptance-criteria placeholders.

The twenty-four column, 2,900 pixel wide grid that crammed fourteen prose fields into narrow boxes is gone. In its place is a compact, scannable and virtualised list - ID, failure mode, use scenario, RPN, residual RPN, status and acceptability - with a full-width form beneath it where the prose fields have room to breathe. Actions now carry due dates and verification-of-effectiveness evidence, with overdue actions counted in the summary bar. Acceptability is derived automatically from residual risk against the document's own thresholds, with manual override retained. Failure modes auto-number and duplicates are surfaced, an acronym glossary explains every abbreviation in the information flyout, and newly added use errors can be synced across from the linked URRA on an append-only basis that never clobbers your edits - now carrying severity and use-error type with them, where previously the scoring was simply dropped.

Decision ladder nodes are movable at last, with positions saved per scenario and a Reset Layout to restore the canonical template; long link labels truncate and expand on selection instead of becoming a wall of text riding the line, and parallel shortcuts fan out rather than stacking on top of one another. Links can be drawn directly on the canvas with their behaviour segments, and scenarios can now be renamed, deleted and duplicated. SOCA gains chip-based allocation cells whose per-cell notes persist as you work, editable and reorderable roles and functions, workload-balance bars and an unallocated-functions callout. Competencies are grouped under Observe, Evaluate and Act bands with quick-classify chips on each row and a classification progress bar, and SRK ratings can be copied between scenarios. Word and PDF reports now embed the abstraction hierarchy and every decision ladder as rendered diagrams.

Loops now keep their number for life rather than being re-ranked as the model grows, appear in the registry within a second of being drawn without pressing Analyse, and carry visible R and B badges on the diagram itself. The simulation engine moves onto the standard System Dynamics effect-multiplier formulation, which measures how far a driver has moved from its own normal operating point rather than treating each causal link as a raw additive rate - so balancing loops converge and reinforcing loops grow S-shaped, as the method intends. Optional per-variable reference values and adjustment times are added, existing models load unchanged, the boundary tool is fully wired, and ErgoLoop gains a Word report where it previously had none.

Both instruments come across onto the platform's theme system. NASA-TLX's six dimension cards and SART's three columns now use neutral surfaces that follow the active theme, with one accent per dimension carrying the colour - NASA-TLX across its six, and SART taking its three from the tool's own results palette, so the editor matches the radar and bars on the findings page. Their icons are redrawn as themeable geometry, and the sliders drop the decorative gradient rail in favour of showing the value itself in the dimension's accent. NASA-TLX's weighting stage gains a progress rail listing all fifteen comparisons, each winner's icon filled in its accent and the loser greyed, with any row clickable to revisit that pair and change your mind - where previously there was a count alone. The flip between pairs becomes a true projection. On SART's findings page the classification tint now sits on a banner inside the score card rather than on the card surface, so it reads cleanly on the Classic and Harry shells. The NASA-TLX and SART dimension descriptions are the validated instrument wording and are untouched, staying in canonical English by design: they are a measuring instrument rather than interface copy, and translating them would quietly invalidate them.

Safety Critical Task Analysis has moved from ErgoTools to Methods, under Task Analysis alongside HTA and GOMS, and is renamed simply SCTA - an "Ergo" prefixed tile read as a stray on the method shelf. Existing SCTA sessions, HTA links and saved workspace pins are entirely unaffected and no data migration is required.

Everything used to share a single 48 pixel title bar - the app logo, save, undo and redo, the menus, Back, the tool's name, its ribbon tabs and the session pill, all competing for one strip. The header is now two rows, as Word's is: the title bar keeps the application icons and the centred session name, and a new full-width row beneath it carries Back, the tool's name, File, the ribbon tabs and the tool's own actions. All thirty-one pages across the seventeen ErgoTools wear it, and despite gaining a row the chrome is four pixels shorter than before. The line between the two rows is gone, and in the Classic theme a single gradient is now split across the boundary rather than each row painting the full sweep and reading as two separate bars. A selected tab underlines and goes semibold while quietly reserving its selected width, so its neighbours no longer shift as you move between tabs. The blank space on the second row drags the window.

Thirty method editors gain the ribbon the ErgoTools got in this release: the nine cognitive pages including SART, ATWIT, IWS, DRT, NASA-TLX, GOMS and HTA; the seven physical ones including RULA, REBA, NIOSH, Snook, MAC, ART and RAPP; the seven reliability tools from HEART and SHERPA through CREAM, SPAR-H, TRACEr, SLIM and THERP; the six usability pages across URRA, uFMEA, SUS and Nielsen; and SCTA. Participant, task and notes move out of the page and into a pinned row that cannot scroll away - MAC, ART and RAPP kept theirs inside the scroll, so it vanished the moment you started work. RULA, REBA, NIOSH and Snook stacked two fixed strips, a participant card above a progress card, and both fold into one ribbon. Mode switches become ribbon groups where they belong: CREAM Basic and Extended, SHERPA and SHERPA+, TRACEr Retrospective and Predictive, SPAR-H At-Power and Low Power/Shutdown, GOMS Edit and Diagram. Results and summary pages deliberately keep their score-hero layouts - a ribbon there would push the finding you came to read further down the page in exchange for nothing. Fields spread the full width of the window and compress on a narrow one rather than scrolling off.

The Edit and View menus are retired from every tool page. Neither carried anything you lose: View duplicated the navigation already down the left-hand side, and Edit held Undo and Redo, which sit as icons on the title bar. In their place Home restores the standard ribbon and Help swaps it, and Help is always last in the row - never, as it had drifted into being on some pages, tucked in beside View. Both words are now genuine tabs: on 63 pages Home was a plain menu item and Help a plain button, neither of which could carry a selection underline even in principle, and both now render through exactly the same styling as the tabs beside them. On ErgoLens and ErgoSense the Help button is rebuilt on the same geometry as those tools' own tabs, so it sits square with them. The selection now tracks the ribbon precisely, including when you start a tour from inside the Help pane, and ErgoTempo's summary page gains the Help affordance to match the rest of the family.

Both tools drove their stages from a segmented chip bar sitting in the page body, duplicating in miniature what the header row above it was already for. Setup, Capture, Review, Analysis and Report are now ribbon tabs in the shell header like every other tool's, with the same enabling rules as before - a stage you could not reach yet is still a stage you cannot reach yet - and the old mode bars are deleted rather than left behind. Two deliberate losses worth stating: the tabs are unnumbered, in the Word manner, and the red findings count that rode the Report chip is gone, because shell tabs carry no badges. The count itself still appears on the Report pages.

ErgoLoop, ErgoBow, ErgoTempo, ErgoStrata, ErgoTrace, ErgoLink, ErgoCWA, ErgoACT and ErgoNeuro now all wear the same shell menu, ribbon, three-column workspace and status bar established by ErgoDesign. ErgoBow gains four ribbon tabs that double as its view switch, with hazard and top-event fields, master data, filters and export collected into the ribbon and a properly divided workspace. ErgoTempo gains a Setup and Record ribbon with a live session timer, Pause All and End Session, plus observation metadata for observer, worker and location. ErgoStrata gains a ribbon, a flush canvas and a properties panel that surfaces the analyst name for the first time. ErgoTrace gains a ribbon over its existing stage wizard with the control-structure palette folded in and pages that finally stretch edge to edge.

Icons across the ribbon tools were a mix of three different media - font glyphs, black and white image pairs, and plain text characters - none of which tinted with the theme. Every one has been redrawn as a single hand-authored family: all of ErgoDesign's ribbon, tab strip and equipment tiles, both ErgoGlare ribbons with the sun timeline and zoom HUD, the whole of ErgoTempo, the remaining icons across ErgoLoop, ErgoBow, ErgoTrace, ErgoLink, ErgoCWA and ErgoStrata, and the icons built in code for dialogs, context menus and row actions. Icons now tint correctly on hover, when active and in danger states, which the old mixture could not do - and every icon reads clearly in all three themes.

The Grid view becomes somewhere you can actually work: drag unplaced elements from a tray onto the canvas, where the only previous way to position an element was typing coordinates, with live coordinate and per-link distance readouts as you drag, a live efficiency chip and a snap toggle. A new five-way view switcher spans Diagram, Matrix, Table, Grid and Compare, and everything is now clickable through to everything else - select an element in the list and it highlights in the diagram and grid, tap a link to open its editor, click a table row to jump back to the diagram. The Matrix gains frozen headers, mirror-cell dimming, hover cross-highlighting and the classic total row and column; the Table gains sortable columns; Compare gains its own key findings panel, an explicit best badge and delta bars. Grid view now reads clearly in the light theme, and the Compare view's weighted travel delta column reports accurately.

A contrast pass across both tools brings receptor badges, selected rows, risk-matrix numerals, banners and critical-task pills up to a consistent, readable standard in every theme, and text in address boxes, description fields, panel labels and status chips now sits comfortably in its space. Both tools adopt a shared panel vocabulary of consistent section bands, one chip system, semantic banners and stat tiles, and ErgoBow's properties pane gains an inspector header that anchors what you have selected. The risk matrix scrolls rather than clipping.

Four different Link-to-HTA dialogs are replaced by one styled dialog serving all thirty-five tool home pages, with pre-selection and a Remove link option, and the New Session popup is standardised across ErgoStrata, ErgoTrace and ErgoGlare. A number of features that were fully built but hard to reach are now surfaced, including ErgoLoop's loop search, sort, filter and insights panel. The control set is tightened to what each tool genuinely offers, and the ErgoCWA home page now carries its HTA, issue and requirement links through into the session.

Version 1.3 brought Word export to every method tool. The ErgoTools now follow, and eleven that had no report of any kind produce a full Word document: ErgoLens, ErgoTempo, ErgoLink, ErgoEnsemble, ErgoDesign, ErgoGlare in both its outdoor and indoor forms, ErgoEvac, ErgoBow, ErgoStrata, ErgoTrace and ErgoCWA - joined by new reports for ErgoACT, ErgoLoop and ErgoNeuro. Each follows the same structure as the established method-tool reports, with a cover page, contents, method explanation and references, findings with tables and charts, and recommendations, and each is filed automatically to the Document Registry as compliance evidence. Diagrams and images genuinely embed: ErgoDesign top and side snapshots, ErgoStrata, ErgoTrace and ErgoCWA diagrams, ErgoGlare false-colour and fisheye renders, and ErgoLens capture snapshots.

A single Word document that tells the whole story of a programme: an assurance summary, the Early Human Factors Analysis, the HF Integration Plan, then one appendix per included tool session - each rendered by that tool's own report builder - and a closing register of every analysis with its status. A new wizard step lets you tick exactly which analyses become appendices, grouped by discipline. Provenance is enforced rather than assumed: each included session is hashed at issue, so re-exporting later flags any appendix that has changed since, and a deleted session renders an explicit note rather than silently vanishing. Cover page, contents and figure numbering run continuously across the whole document.

Every report preview in the product was a single continuous column with no pagination whatsoever - page breaks existed only once you had already sent it to a printer, and printing itself was hidden in a right-click menu. A preview now shows what you are about to produce: floating A4 pages on a print canvas, each with the document title as a running head and a page number at the foot, and a toolbar carrying Print and Export DOCX where you can find them. It covers the thirty-five or so tools on the shared report shell, the four with bespoke builders - ErgoTrace, ErgoBow, ErgoCompass and SCTA - and ErgoDesign, which moves off its own private report window onto the shared one. Nothing changes on the Word side: DOCX export already existed for every tool, and the toolbar simply reaches it. The preview is an honestly paginated view of the HTML report rather than a rendering of the Word document, and the two are not claimed to be identical. ErgoDesign is the one tool with no Export DOCX button in its preview, deliberately: its preview draws live from what is on screen while its Word export reads the saved document, and placing both on one toolbar would be an invitation to export something other than what you were looking at.

The Early HF Analysis and the HF Integration Plan now come out of Word looking like every other ErgoSphere report: title page, contents and the standard 25.4 millimetre margin, across all seven of their export flows. Their content is fuller for it - the EHFA now renders its version-pinned upstream citations, an HFIP with an empty schedule says plainly that nothing is scheduled, and both build their front matter properly. Both gain a standalone structure - about, how to read it, document control, executive summary, references and appendices - with the HFIP carrying its provenance legend and complexity bands, and the compiled closeout gaining a preamble that explains the three content states, the five assurance questions, the difference between a gap and a caveat, and what a valid, stale or absent sign-off means. Where the document offers writing guidance it lives inside Word's placeholder controls, greyed and clearing the moment you type, so template prompts can never survive into a finished report. All eight export flows now open the document once they have written it. Both documents also state the competence expected of the practitioner behind them: a post-graduate qualification in ergonomics or human factors and membership of a recognised professional body, HFESA or CIEHF, maintained through CPD.

The Compliance module is renamed Assurance, and PDF is fully retired from it - every report surface is now Word or an in-app HTML preview. Six register reports export to Word with wide tables landing correctly on landscape pages: Compliance Matrix, Applicability Statement, Change History, Professional Judgement, Executive Summary and Sign-Off Register. Evidence packages now bundle self-contained HTML reports that open in any browser. Three new assurance document cards - Early HF Analysis, HF Integration Plan and HF Closeout - carry live status and one-click Word export, and a small programme can now run a closeout without an HF Integration Plan, with the report saying so honestly rather than reporting zeros.

A licence's contents were previously baked in at activation, so changing a customer's entitlement centrally did not reach an already-activated machine and the licence key had to be re-entered by hand. Entitlements now refresh silently on the routine online check: an upgrade or a change of plan simply arrives, with no re-activation and no customer action. The safety properties are deliberate - an expiry date can only ever be extended by a refresh and never shortened, so a renewal reaches you without a re-key while a routine check can never silently shorten a working licence; revocation and genuine expiry are checked before any refresh, so a refresh can never resurrect a revoked licence; and a malformed refresh is ignored in favour of the working licence you already hold.

A memory audit across the platform keeps long working sessions running as smoothly at the end of the day as at the start. Nine diagram pages - ErgoCWA, ErgoTrace, THERP event trees, ErgoStrata, HTA, GOMS, ErgoBow, ErgoLoop and ErgoLink - now release their embedded browser controls when you navigate away, which was the single largest win of the audit. Alongside it: the notification panel subscribes once rather than on every load, ErgoSense detail charts release their drawing resources on repaint and zoom, and the caches behind the ErgoGlare sun-path service and the compliance reports page are now bounded.

A new save coordinator serialises every path that can write to your project - manual save, Save As, timed auto-save, crash recovery, restore and auto-save cleanup - so each completes in a defined order and an auto-save can never overtake the save you asked for. Emergency saving during an unexpected shutdown is bounded to two seconds rather than waiting indefinitely, so the application always closes cleanly.

A verification pass across the platform confirms that every figure on screen traces back to a tested calculation. ErgoNeuro's five band powers and its cognitive load index come from its unit-tested signal processor and from nowhere else, and the tool says so explicitly when signal quality is too poor for those numbers to be relied upon. ErgoEvac's analysis figures are computed from the simulation itself, and its report presents Egress Performance Observations alongside a plain statement that the model does not assess code compliance - that judgement belongs to a certifier, not to a simulator, and the report does not imply otherwise.

An in-progress draft in the EHFA, HFIP or Closeout wizard is picked up where you left it after a restart. Finalising writes the document before committing the frozen record, so a file left open in Word cannot leave you with a finalised record that was never exported, and every frozen artefact can be re-exported later from its row. Re-finalising an EHFA mints a second version alongside the first, preserving the record that earlier registry entries and plan seeds already cite, and finalising now requires an EHFA with content in it. ErgoEvac's CSV export reads correctly on French, German and Dutch system locales.

Despite the scale of this release, it is additive. Existing projects and sessions open cleanly: URRA files migrate to the new two-level task and error structure automatically with no data loss, uFMEA rows keep their recorded acceptability verbatim, ErgoLoop models load unchanged and fall back to their existing values where the new reference settings are absent, and SCTA sessions, HTA links and workspace pins are untouched by the move to Task Analysis. Where a refinement genuinely shifts results - occupants no longer reading exit signs from behind in ErgoEvac, ErgoLoop's replaced simulation mathematics, or ErgoLens's rebuilt shoulder and trunk measurement, which sharpens RULA and REBA scoring from video and photo capture - it is stated in the tool and recorded in the report's assumptions.

ErgoACT, TRACEr, ErgoNeuro, ErgoEvac, ErgoCompass and ErgoAIR ship as Preview tools in this release. They are complete and usable, and their methods and calculations are sound, but they are new: their interfaces and outputs may evolve in response to what practitioners tell us. ErgoNeuro has been verified end to end on its demo signal path; hardware headset paths are provided but not yet certified. ErgoAIR is the newest of them and was designed ahead of the operator workshops it is built to serve, so its worksheet columns and priority bands in particular are the ones we most expect to change once practitioners have run real sessions through it.