We build societies in software,
then break them
on purpose.

Faultline stands up independent actors, applies a real decision, and shows you exactly where the coalition cracks, before you commit to it.

Request a briefing → See how it works

400+ agents · 7 cascading layers · 5 analysis modes

The mechanism

Each actor in the simulation reasons independently.

Most simulations are weighted averages. Ours are not. Each actor responds in character, then is held accountable to its own record across runs.

Identity & incentives

Who they are, what they stand to gain or lose. Weighted ties to allies and rivals, carried into every run.

Documented behaviour

Prior positions and public record inform how each actor reacts. Reactions that contradict it are flagged.

Grounded in current context

Each run starts from what actually happened, not a hypothetical. Behavioral traits are continuous, not fixed labels.

The cascade

A stimulus enters. The whole system reacts.

Every actor in the simulation (institutions, leaders, stakeholder groups, amplifiers) responds to what the others actually decided, not to an assumption. Allies amplify, rivals dampen. Consequences build the way a real system would.

Institutions & leadership

Organizations and the people accountable for them, reasoning from their actual track record, not a generic persona.

Stakeholder groups

Populations and interest groups with their own incentives, loyalties, and material stakes in the outcome.

Amplifiers

The press, platforms, and narrative channels that decide how far a signal travels.

Ask any what-if.

Change the timing, the messenger, or the sequence. Run it again and compare.

What if we announce this on Thursday instead of Monday?
What if a key ally breaks ranks first?
What if the opposition gets ahead of the story?
What if we walk it back after initial pushback?

What the report covers.

Fault lines, drift signals, and anomalies, each one tied to the actors and evidence behind it, not a bare summary.

Analysis modes

Five ways to put a question to the system.

Cascade

How does the whole system react, and where does it crack?

Council

Would an expert panel approve, revise, or kill this before it ships?

Adversarial Test

Auto-played adversarial exchange: best counter, then counter-to-the-counter.

Exposure Scan

Automated red-team: generate attacks on an actor and rank real exposures.

Structural Read

Instant structural read: group exposure and what-ifs, no simulation run.

Why it's defensible

Generating reactions is easy. Trusting them is harder.

Evidence Receipts

Every output carries its source. No claim without provenance.

Belief Alignment

Reactions that contradict an actor's documented record are caught before they reach you.

Self-Consistency

Statistical outliers are flagged automatically before they land in the report.

Deterministic Floor

Structural reads derived in pure arithmetic. Identical on every run.

Where it runs

Started in politics. Built for anything that cascades.

The people and stakes change. The rigor doesn't.

Electoral intelligence

Live

The most granular electoral simulation running in India today. Submit a decision. Get a read of where it lands, who it moves, and where it fractures, before the campaign commits.

Tamil Nadu · 234 constituencies · full simulation coverage

See how it works →

Corporate policy

Model how a major decision ripples through an organization before it ships. Who resists, who amplifies, and where it stalls.

Business units · employees · regulators · investors · press

Defense & geopolitical

Planned

Multi-actor escalation, alliance strain, adversary reactions, mapped before the move is made.

States · factions · commands · domestic constituencies

Brand & crisis