Hyveminds/Products/SimChoices
Product 01 · Spotlight · v3.4 · 2026 cohort

Branching
sims from real
practice.

A canvas for the messy, conditional, “it depends” conversations that don’t fit a multiple-choice quiz.

SimChoices turns expert practitioners’ war stories into branching scenario simulations, with NLP that reads what learners actually wrote, path validators that catch dead-ends before launch, and SCORM/LTI 1.3 hand-off into your LMS. Built with TAFE NSW, Western Sydney University, and the Per Capita teaching network.

02 · What it is

A canvas for conditional teaching.

Most digital learning collapses real practice into a multiple-choice quiz. SimChoices keeps the conditional logic intact, what an expert actually does when context, history and consequence all matter at once.

For authors

Capture the story.

Sit beside a senior practitioner. Record their case. The canvas guides you to break it into scenes, decisions and consequences, the same shape as the way they tell the story over coffee. Story stays the unit of design; the engine handles the branching.

For learners

Type a real response.

No multiple-choice unless the scene calls for it. Learners write what they would say, draft what they would send, plan what they would do next. NLP reads it, classifies the approach, and routes the simulation to the consequence that follows.

For programs

See the field.

Cohort analytics show where the path forks, not just whether learners passed, but which approach they took and how it changed under pressure. Hand off to your LMS via SCORM 1.2 or LTI 1.3 with grades intact.

03 · A walk through

Three views of the same sim.

The same scenario seen by the learner taking it, the educator who built it, and the program leader watching the cohort. Drop into any tab to see the surface area.

simchoices.app/play/early-childhood/disclosure-conversation
Scene 4 of 7 · Disclosure conversation · Early Childhood
M
Maya, 4, at pickup, quietly

“My mum’s boyfriend says I’m not allowed to tell anyone, but… he keeps coming into my room at night.

What do you say next? Type your response or pick a starting line.
State
Trust score+2
Disclosure stageopen
Time pressure3:42 left
Witnessesnone
Reference
CHCPRT001, Identify and respond to children and young people at risk · §3.2 procedural fairness in disclosure.
Engine
Pathresponsive · open
NLP confidence0.91
Next scene05 · seek
simchoices.app/author/early-childhood/disclosure-conversation/canvas
Start · Scene 01 Maya approaches at pickup
Scene 02 · Open “Tell me more” · responsive
Scene 02 · Verify “Are you sure?” · doubting
Scene 02 · Escalate “I’ll tell supervisor”
Scene 03 Disclosure deepens · seek
Scene 03 Maya withdraws
End · feedback Procedural-fairness debrief
Scene 04 Document & report
End CHCPRT001 §3.2 reflection
Validator · 0 issues
9 nodes · 12 edgesVALID
simchoices.app/programs/tafensw-cs/cohort/2026-march/analytics
Path distribution · Scene 02
312 learners · 96% complete
Open · responsive62%
Verify · doubting21%
Escalate prematurely14%
Other / unparsed3%
Reflection delta
pre vs. post · self-rated confidence
DISCLOSE
DOC
REPORT
REFLECT
DEBRIEF
NLP heard · sample of learner responses
classified into archetypes · auditable on click
“Thank you for telling me, that took real courage. Can you help me understand what happens?” responsive · 0.94
“Maya I hear you. I’m going to write down exactly what you said and we’ll find someone who can help.” responsive · 0.88
“Are you sure that’s really what’s happening? Sometimes things look different at night.” doubting · 0.82
“I need you to wait here while I get my supervisor. Don’t move.” escalate · 0.79
04 · Capabilities

What ships in the box.

Nine capabilities that turn the canvas from a slide-deck replacement into a research instrument. All available on every plan; pilots get the full surface area at no cost.

01 · Authoring

Branching canvas

Drag-and-drop graph of scenes, choices and consequences. Up to 7 layers deep, with pattern-locked sub-trees so re-using a sub-flow stays in sync.

02 · Engine

NLP response classifier

Open-text learner replies are classified into author-defined archetypes via a tuned BERT pipeline. Auditable confidence scores; manual override always one click away.

03 · QA

Path validator

Catches dead-ends, unreachable scenes, dangling branches and unused choices before publish. Walks every reachable path, surfaces the gaps with a one-click jump.

04 · Runtime

Stateful timing & pressure

Scenes can be timed, paced or interrupted. Time-pressure variables flow through state and influence which path opens next. Useful for triage and crisis sims.

05 · Integration

SCORM 1.2 & LTI 1.3

Hand off into Canvas, Moodle, Blackboard, Bridge, Brightspace and any LTI 1.3-compliant LMS. Grades, completion and time-on-task pass back automatically.

06 · Analytics

Cohort path analytics

See where a cohort branches, where they get stuck, where confidence climbs. Filter by demographic, repeat attempt, or facilitator. Export to CSV or SPSS.

07 · Accessibility

WCAG 2.2 AA, audio paths

Every scene supports a recorded-audio variant; transcripts, captions, keyboard nav and reduced-motion paths shipped on day one. Vetted with Vision Australia, 2024.

08 · Authoring AI

Story-to-scenes assistant

Paste a transcript of an expert telling their case. The assistant proposes scenes, decision points and archetypes, you accept, edit or reject each one before it lands on the canvas.

09 · Research

Built with educators

Every release is shaped with real teachers and trainers, validation cohort size, classifier F1, inter-rater reliability, drop-off curves all reviewed with the people who use it.

05 · Workflow

Story to cohort in five steps.

How a sim moves from a senior practitioner’s anecdote to 312 learners in a TAFE cohort.

01 · Capture
i

Sit, listen, record.

An author runs a 45-minute conversation with a senior practitioner. The story-to-scenes assistant turns the transcript into a draft tree of scenes and decision points.

02 · Author
ii

Lay out the canvas.

Edit nodes, write archetypes for the open-text choices, set state variables (trust, time, escalation). Validate. Iterate. The graph stays the source of truth.

03 · Pilot
iii

Run with 30 first.

Soft-launch to a small cohort. The dashboard surfaces unparsed responses; you tighten classifier archetypes until F1 climbs above 0.85.

04 · Deploy
iv

Hand off to the LMS.

Publish to SCORM 1.2 or LTI 1.3 endpoint. Grades and completion pass back automatically. Scale to a full cohort with a single configuration change.

05 · Reflect
v

Read the field.

Cohort analytics show what the field actually did, not just whether they passed. Iterate the canvas based on where the path forks and stalls.