EMERGENCY RESCUE RAPID DECISION SUPPORT SYSTEM · ALAYA RESCUE LAB

In the disaster operations room, the AI picks the tools; people make the call.

Forest fire, a missing hiker, a fall into a river, roads to reopen, a lost link: the computation these need is the same set of engines as in the Tactical Lab. Here it is rerun in rescue terms. The upper layer is ZERO turning a request into a computable question; the lower layer is the tools it calls, each one open to try yourself.

All simulation
Data stays in country
People stay accountable (ICS unchanged)

01 ZERO picks its own tools 02 The tools it can call 03 Six rescue scenarios 04 Accountability and limits

ZERO picks its own tools

The language model, deployable on-prem, does one thing: it turns a sentence into a computable question and lays the answer out with its sources.

Dialogue: describe the situation and the limits in a sentence
Intelligence de-duplicated, with sources: one event once, every line sourced
Three courses of action: each with numbers and assumptions
Human approval: the commander chooses; nothing is decided for them
Execution plan: dispatch and reports, recomputed on every change

ReproducibleToday's ZERO already calls the terrain, line-of-sight, radio-horizon and ballistics engines itself and returns the numbers with their sources.
WiredZERO's reconnaissance requests and the execution plan the commander approved go straight into the ALAYA operator console queue, where a person acknowledges, reports or rejects; both sides see the same state.

The tools it can call

Every card is a page that really computes; the ones marked “by arrangement” run on ZERO's own host.

01By arrangementReproducible

Terrain and line of sight

What can be seen from where: observation points and drone altitude for a mountain search.

02By arrangementReproducible

Radio horizon

Reach of relays and robot dogs, and which valley stays silent.

03Research accountReproducible

Probability × value search

Where to fly first with limited sorties: the search console for wildfire survivors, people in the water and rafts.

04No loginReproducible

Road repair ordering

On the real roads of Jianshi: which cut first, and when each village gets its road back.

05No loginReproducible

Mission continuity on link loss

What the aircraft do when the link drops: the swarm-autonomy demonstration.

06Research accountReproducible

Drift model

Where a person and a raft drift: the drift and survival weighting inside the sea mission.

07By arrangementReproducible

ZERO rescue workbench

Start from one road-closure alert and watch ZERO pick its tools and propose.

Six rescue scenarios

Each scenario lists the tools ZERO would call; where there are numbers, they are published simulation results against the conventional method on the same scenarios.

Research accountReproducible

Wildfire: hot spots and trapped people →

Where to fly first with limited sorties: new hot spots and trapped people around the fire perimeter, weighted by assets and wind.

	ALAYA weighted search	Conventional
Trapped people found	93%	65%
Hot spots found	73%	62%
Hot spots near key assets	83%	66%

Numbers: 200 synthetic scenarios, assuming the intelligence is correct; compared with the conventional flight pattern on the same scenarios.

ZERO callsProbability × value searchTerrain and line of sightRadio horizon

No loginIn preparation

Missing person in the mountains

The Baoshan type of case: last known position, trails and sightlines, which ridge first. The mountain search view is in preparation; the method is the one used for wildfire and sea.

The same method is shown today on the wildfire and sea missions.

ZERO callsTerrain and line of sightProbability × value searchRadio horizon

Research accountReproducible

River and near-shore rescue →

A person and a raft drift differently and survival time is counting down: drift model plus survival weighting, which cell first.

	ALAYA weighted search	Conventional
Found overall	72%	53%
People in the water	74%	57%
Rafts	71%	50%
Nothing found at all	6%	21%

Numbers: 200 synthetic scenarios, assuming the intelligence is correct; compared with the conventional flight pattern on the same scenarios.

Maritime SAR brief (four drones from a cargo ship) →

ZERO callsDrift modelProbability × value searchMission continuity on link loss

No loginReproducible

Road reopening and cut-off villages →

A typhoon cuts the mountain roads in several places: which cut first, when each village is back, against the whiteboard rules. Computed on the real roads of Jianshi.

ZERO callsRoad repair orderingRadio horizonTerrain and line of sight

No loginReproducible

Mission continuity when the link drops →

The link drops in a valley: how the search goes on, who takes over, what gets reported — the swarm-autonomy demonstration.

ZERO callsMission continuity on link lossRadio horizon

No loginReproducible

Perimeter watch at a major facility incident →

A petrochemical or port incident: layered perimeter watch and a human-approved response ladder, adaptable to the rescue setting.

Currently the defence version of the screen; adaptable.

ZERO callsRadio horizonTerrain and line of sightMission continuity on link loss

Accountability and limits

Simulationevery number here comes from synthetic scenarios, not from records of a real disaster.
Data stays in countryin a production deployment the model and the data sit on premise or in a designated facility and pass through no foreign service; everything shown here is synthetic.
People stay accountablethe incident command system is unchanged; the AI supplies options and evidence, and dispatch is approved by a person.

Want to see it in your own operations room?

Bring a real situation; we rerun the same engines on your data and watch, on the spot, which tools it calls and what it proposes.

Request a private demo Back to the Tactical Lab