Build log

Changelog

What we've shipped — the engine, the scenarios, and the tools behind them.

22 Feb 2026

Engine

The Flight Recorder: Deterministic Replay & Bug Reconstruction

A bug that happens at minute 14 of a chaotic scenario is a nightmare to debug — unless you can play it back like a movie. We built a deterministic replay engine that captures, shares, and reconstructs any simulation session with bit-perfect accuracy.

A JSON history log being replayed through the simulation UI with a playback slider

As we added more concurrent actors — patients, doctors, room timers — we hit the classic distributed systems problem: a bug would appear once in a session and vanish when we tried to recreate it.

The fix was architectural. Every session is now a pure function of its inputs, recorded as an event stream:

  • Both the simulation heartbeat and every user command are recorded with high-resolution timestamps, maintaining perfect causality on playback.
  • Random ID generation replaced with a deterministic system — every room and patient has the same identity across every replay.
  • Each session produces a portable JSON flight log that can be exported and replayed to jump to any moment in the scenario.

For the simulation, this lays the groundwork for post-drill debriefs — scrubbing back through decisions, seeing exactly where a treatment plan succeeded or failed.

8 Feb 2026

Simulation

Dynamic Orders & Staff Tracking

Managing a mass casualty event requires more than clinical knowledge — it requires the ability to sequence interventions under time pressure. This update adds real-time order prioritization and staff management across the full patient trajectory.

The room drawer view showing dynamic order prioritization and staff assignments

Players can now manually re-order medical commands in real time — deciding between a 1-minute life-saving intervention and a 25-minute diagnostic procedure, across multiple patients simultaneously.

A dedicated staff tracking system differentiates between specialists, residents, and nursing staff, and visualizes availability to prevent bottlenecks during critical arrival waves.

The simulation has expanded beyond the shock room. The architecture now supports a complete patient trajectory including triage, care units, imaging, and surgery.

1 Feb 2026

UI

Semantic Activity Icons & Live Tooltips

Generic progress bars tell you something is happening. In a mass casualty scenario, that's not enough. We replaced them with activity-specific icons that pull live metadata directly from the patient's state machine.

Before and after: generic progress indicators vs. activity-specific icons with tooltips

A lab beaker signals blood work. A dedicated icon for surgery distinguishes it from stabilization. Hovering over a staff member now shows exactly what they’re doing and for whom — “Oxygen — Mr. John Doe.”

Icons are mirrored across patient and staff cards, so a coordinator can read the state of the room at a glance without clicking into individual records.

17 Jan 2026

Scenarios

The Scenario Library

The emergency department never sees the same day twice. We overhauled the core architecture to support multiple distinct crisis scenarios — different events, different patient loads, different environmental conditions.

The scenario library showing multiple selectable crisis events

The first new scenario: Flight 100, an aviation disaster. Patients arrive via ambulance on scenario-specific timelines rather than starting in the ED. Pre-hospital data — ambulance handovers, field interventions — is accessible before the first patient hits the door.

Scheduled situational messages provide evolving data on casualty counts, hazards, and arrival waves in real time.

7 Jan 2025

Founding

Why we started building Youfers

Paper-card MCI drills produce data you can’t trust. Someone follows each player with a stopwatch. At the end you get a report that says your team handled 14 patients — including CT imaging for all of them — in two hours. A CT scanner processes one patient at a time. The math doesn’t work, but the manual tracking can’t tell the difference.

We started building Youfers to fix that. A simulation that models your actual department — your resources, your staff, your constraints — and tracks every decision automatically. This changelog is the record of what we’ve built since.