.png)
Docs / The Agents
The Agents
FluxForce runs 28 specialized AI agents. Each has a job, a name, and a clear remit. They're grouped into three squads, and they share what they see in real time.
Why a team, not one model
A single model trying to do fraud, sanctions, identity, and governance at once is a model that does none of them well. FluxForce splits the work the way a real institution does: specialists who are good at one thing, talking to each other. Each agent carries a clear designation, Senior, Lead, or Director, that signals how much weight its decisions carry.
The three squads
Threat
Catches what's actively trying to hurt you: fraud, money laundering, account takeover, and suspicious movement of funds. The squad most teams switch on first.
Trust & Identity
Answers "is this person who they claim to be." Onboarding checks, identity verification, sanctions and watchlist screening, and ongoing customer due diligence.
Risk & Governance
Watches the watchers. Customer risk scoring, regulatory reporting, model oversight, and the evidence that keeps every decision defensible.
A few of the team
The full roster covers 28 agents. Here are a few you'll meet early, with their current availability.
| Agent | Squad | What they do | Status |
|---|---|---|---|
| Aiden Flux | Threat | Fraud detection across transactions and accounts. The first agent most teams put to work. | LIVE |
| Nova Sentinel | Threat | Watches for money-laundering patterns and suspicious movement of funds. | BETA |
| Iris | Trust & Identity | Identity verification and onboarding checks at the point of entry. | BETA |
| Leo | Trust & Identity | Sanctions and watchlist screening, run as a hard check before any scoring. | BETA |
| Zara | Risk & Governance | Customer risk scoring and ongoing due diligence as behavior changes. | BETA |
More agents come online across the three squads on a rolling basis. Availability shown above is current at time of writing. Your account team can share the live roster for your deployment.
How agents work together
They self-activate on signals
Agents aren't dispatched by a central controller. When a signal appears that's relevant to an agent, it wakes up and goes to work. No bottleneck, no single point of failure.
They share what they see
A fraud signal from one agent is visible to the identity and risk agents the moment it lands. Weak signals in several places combine into a strong one.
They check each other on big calls
High-impact decisions aren't left to a single agent. More than one weighs in and has to agree before anything happens. Consensus, not solo authority.
Each agent runs at the autonomy level you set, and any agent can be paused on its own without touching the rest. See Autonomy & Controls.
FAQ
Do we get all 28 on day one?
You can, but most teams don't. Start with the agents that matter most, usually fraud and sanctions, and add the rest as you build trust and connect more data.
Can agents be turned off individually?
Yes. Autonomy and on/off are per agent. Pausing one doesn't affect the others.
Are these the same agents across industries?
The squad structure and the operating system stay the same. The agent library is tuned per industry. Today's library is built for financial institutions.