02/Capabilities
Two layers, modelled honestly.
L1/The layers
The physical layer. Geometry, sightlines, and the clock.
The measurable structure of a place and a plan: where exposure opens, where it narrows, and how long each state holds. Five readings, each on its own terms.
L1.1/Sightlines
Where a principal can be seen.
We trace where a position is visible, from where, and through which windows of movement. Sightlines open and close as people move, as a vehicle repositions, as a door is held. We model them as a function of the plan.
L1.2/Chokepoints
Where flow narrows and time rises.
Arrivals, thresholds and transfers thin the plan to a single lane. Time on target climbs exactly where the plan has least slack. We find where many paths converge on one gate, and what it costs when that gate is contested.
L1.3/Exposure windows
Exposure measured against the clock.
Exposure is a duration. We measure how long a position holds before it changes, where the peak sits, and which moments carry the most risk. A window you can measure, you can shorten.
L1.4/Comms coverage
Where the radio plan thins.
Coverage overlaps in some places and thins in others, and the gaps rarely sit where anyone expects. We map where the comms plan holds, where it doubles up, and where a dead zone lands on a moment the plan cannot afford to lose contact.
L1.5/Timing
The plan as a sequence.
A plan reads differently as a timeline than as a diagram. We lay it out as a sequence, show what runs early, what runs late, and what cascades when one step slips, then surface the critical path where a small delay becomes a large one.
L2/The layers
The behavioural layer. How people act under pressure.
Geometry alone is inert. We populate the model with tens of thousands of independent AI agents, crowds, staff and, where engaged, adversary agents, each making its own decisions, then apply the pressure that tests whether a plan that looks sound on paper survives contact.
L2.1/Crowds
How crowds form, flow and surge.
Simulated crowds form, flow, surge and disperse around the plan. Density builds toward thresholds and breaks without warning. We watch where it concentrates and test what the plan assumes about a crowd that never read it.
L2.2/Staff
How posts and teams respond.
Posts, teams and stewards respond to instruction, fatigue and surprise. We model how the human side of the plan behaves when an incident pulls attention, when a position runs thin, or when two orders reach the same person at once.
L2.3/Adversary profiles
Modelled pressure, where engaged.
Where an engagement calls for it, we introduce modelled adversary behaviour, abstract profiles and never real people, to probe the plan from the outside. The purpose is defensive. We find where a plan is soft so you can make it firm.
L2.4/Pressure
The same plan, under load.
Everything above shifts the moment an incident, diversion or comms loss enters. Pressure is where the two layers meet. We test geometry and behaviour together and run them until the plan's response under load is understood rather than assumed.
↳/The engagement output
The vulnerability-and-timing report.
The two layers resolve into one deliverable: a report, returned through a channel you control. It hands you the places to look, ranked by how much they matter and how long they last.
The report supports your planning and your exercises. It informs decisions and leaves them with you. The judgement, and the plan, stay yours.
- —Exposure. Where, and for how long, measured against the clock.
- —Single points of failure. Where one chokepoint, post, channel or decision carries too much.
- —Where it breaks. Where route, message or comms give under load, and which perturbations push them there.
Make contact
Start a confidential conversation.
Engagements are vetted and selective. The first step is a short private exchange under NDA. There is no platform to sign up for.