.png)
Docs / Getting Started
Getting Started
From a signed order form to your first live risk signals in days. Here's the path, and what your team handles versus what we handle.
Before you start
A short list keeps the first week clean. You'll move faster if these are ready:
- A named owner on your side, usually someone in fraud, AML, or compliance ops.
- Read access to the systems you want FluxForce to watch first. Most teams start with one: payments or transaction monitoring.
- Your deployment choice. SaaS to start fast, on-premise to keep data inside your walls, or hybrid.
- A rough sense of which decisions you'd let an agent make on its own, and which always need a human. You can change this later.
The five steps
Create your workspace
We provision a tenant in the deployment model you picked. Your team gets accounts, roles, and a private space. Nothing is shared with other customers.
Connect your first source
Point FluxForce at one system to begin, usually transaction monitoring or payments. Sensitive fields are tokenized as they arrive, so raw customer data isn't sitting in the clear. See Connecting Data Sources.
Turn on a starter squad
Begin with a small set of agents rather than all 28. Fraud detection is the usual first one. The agents start reading your live data and building a picture of normal behavior.
Set the autonomy line
For each active agent, decide what it can do alone and what waits for sign-off. Start cautious. Most teams run in shadow mode first, where agents recommend but don't act. See Autonomy & Controls.
Review the first decisions
Within days you'll have real signals and real decisions, each with its reasoning attached. Work through them with your team, adjust thresholds, and widen autonomy as trust grows.
You don't have to hand over control on day one. In shadow mode every agent makes its call and records the reasoning, but takes no action. You compare its decisions against your team's for a few weeks, then switch on autonomy where the agent has earned it.
What "first value" looks like
Value isn't a finished rollout. It's the first time an agent catches something your current process missed, or clears a false positive your analysts would have spent an hour on, and shows its work. That usually happens in the first couple of weeks on a single data source. Breadth comes after.
FAQ
How long until we see something real?
Days, not quarters, for the first source. A full multi-source rollout takes longer and depends on how many systems you connect and how fast access is approved on your side.
Do we have to connect everything at once?
No. Start with one system. The agents work with what they can see and get sharper as you add sources.
Can we keep data on our own infrastructure?
Yes. On-premise and hybrid deployments keep sensitive customer data inside your environment. See Connecting Data Sources.
What if an agent gets something wrong early on?
That's what shadow mode is for. Nothing acts until you've seen how it decides. And the kill switch is always available once you do switch autonomy on.