FluxForce

Docs / Connecting Data Sources

Connecting Data Sources

FluxForce reads from the systems you already run. You don't move your data into a new home or rip anything out. You point us at a source, and sensitive fields get tokenized before anything downstream sees them.

What FluxForce can read

Most financial institutions start with one source and grow from there. Common ones:

  • Core banking and ledgers. Accounts, balances, and the record of what moved where.
  • Payments and transaction monitoring. The live stream most fraud and AML work depends on.
  • CRM and onboarding systems. Who the customer is and how the relationship started.
  • Case and ticketing tools. Past investigations and their outcomes, which the agents learn from.
  • External intelligence. Sanctions lists, watchlists, and other outside signals that inform a decision.

Sensitive data is tokenized on the way in

This isn't a downstream cleanup step. The moment data enters FluxForce, fields like names, account numbers, and identifiers are tokenized. The agents reason over tokens, not raw personal data. So a decision can be made, explained, and stored without raw customer details sitting exposed inside the system.

Your data can stay on your side

On-premise and hybrid deployments keep sensitive data inside your own environment. FluxForce runs where your data lives instead of pulling it out to a place you don't control.

How a connection is set up

1

Pick the source and the access

Choose the system and the read access FluxForce needs. Read-only to start is fine, and common.

2

Connect

Use a supported connector, an API, or a secure feed, depending on the system. See Integrations for what's supported today.

3

Map and tokenize

We map the fields the agents need and set tokenization rules for the sensitive ones. This is where you decide what stays masked.

4

Verify, then go live

We confirm the data is arriving cleanly and the tokenization holds, then the connected agents start reading the live feed.

What we don't do with your data

  • We don't train a shared model on your customer data. Your tenant is yours.
  • We don't need write access to start. Many teams run read-only for weeks.
  • We don't keep raw sensitive fields lying around once they're tokenized.

FAQ

Do we have to give write access?

Not to begin with. Read-only is enough for shadow mode and for most monitoring. Write access matters only when you want an agent to act directly in a source system, and even then it's scoped.

What if a system we use isn't listed?

Connectors cover the common ones. For anything else there's an API and secure feed option. See Integrations.

Where does tokenization happen?

At ingestion, before any agent or model sees the data. It's the first thing that happens, not the last.