.png)
Docs / How It Works
How FluxForce turns data into defensible decisions
Plenty of tools tell you something looks wrong. FluxForce works out what to do, acts inside the limits you set, and keeps a record an auditor can follow. Here's the path a single event takes.
The problem we solve
Compliance and fraud teams live with three gaps. Alerts pile up faster than analysts can clear them. The systems that raise those alerts don't talk to each other, so a fraud signal and a sanctions hit on the same customer never meet. And when a regulator asks why a decision was made, the answer lives in someone's memory or a spreadsheet. FluxForce closes all three.
The path of one event
Read
An event arrives, a payment, a login, a new account, a wire. FluxForce ingests it from your connected systems. Sensitive fields are tokenized at the point of entry, before anything downstream touches them.
Understand the context
Before any scoring, the event is enriched with what's already known: the customer's history, related entities, prior cases, and outside intelligence. Context comes first, so a decision isn't made on a single transaction in isolation.
Apply hard rules, then judgment
Non-negotiable checks run first. A sanctions match stops the line before any model weighs in. Only after the hard constraints pass does the scoring step assess how risky the event is and why.
Decide and act
The responsible agent reaches a decision and takes the next step within its autonomy setting: clear it, hold it, raise a case, or escalate to a person. Low-risk actions can run on their own. High-risk ones always wait for a human.
Record the reasoning
The inputs, the rules that fired, the model's output, the final decision, and any human sign-off are written to a tamper-proof record. This is the part that holds up months later. See Evidence & Audit.
Learn from the outcome
When a case is confirmed or cleared, that result feeds back. The agents get sharper on your institution's real patterns over time, not on a generic benchmark.
Agents reason together, not in silos
The agents don't work alone. When the fraud agent sees something, that signal reaches the sanctions and customer-risk agents in real time. A weak signal in three places can add up to a strong one. For decisions that carry real weight, more than one agent weighs in before anything happens, so no single agent has unchecked authority over a high-impact call.
A risk score on its own isn't an explanation. FluxForce records the full reasoning chain: which signals mattered, which rules applied, and how the agent weighed them. If you can't explain it to a regulator, the decision isn't finished.
What stays under your control
- The autonomy line. You set, per agent, what runs alone and what waits for a person. See Autonomy & Controls.
- Where data lives. SaaS, on-premise, or hybrid. Sensitive data can stay inside your environment.
- The off switch. One control pauses autonomous action across any agent, any squad, or the whole system.
FAQ
Is this a dashboard product?
No. Dashboards show you data and leave the decision to you. FluxForce makes the decision, acts on it within your limits, and shows the reasoning. The dashboards are a view onto that, not the point.
How is it different from a rules engine?
Hard rules still run, and they run first. But rules alone can't read context or learn from outcomes. The agents add judgment on top of the rules, and they explain how they reached it.
Does it replace our analysts?
It takes the triage off them. The routine clears get cleared, the obvious holds get held, and the genuinely hard cases land on an analyst's desk with the context already gathered.
How do we trust a decision we didn't make?
Start in shadow mode and compare. Once an agent's calls match your team's, switch on autonomy for that agent. And every decision it makes carries its reasoning, so trust is checkable, not assumed.