%20(1).jpg)
Routing has been solved by multiple companies in many different ways. Orchestrators can send a transaction to the right acquirer based on a set of rules. Easy.
The harder problem, and the one the industry rarely talks about, is making that logic safe, flexible and understandable to the people who actually manage it.
That’s the problem we set out to solve at Paytently.
The cost of getting it wrong
A misconfigured routing rule obviously degrades conversion, but can also lose revenue, create compliance exposure, or silently send traffic down the wrong path for days before anyone notices. The real risk is human configuration mistakes in a system too complex to reason about clearly.
Most platforms respond to this in one of two ways. They either oversimplify by offering a handful of preset options that limit what payment teams can actually do, or they expose the full system complexity and hope the user gets it right. Neither approach respects both the problem and the person solving it.
A simple model for complex logic
We took a different path. Paytently’s routing engine can route on any data point included or enriched in a payment object, making it genuinely flexible and extensible. However, the system is based on a true/false decision tree, visually translated into a flowchart UI that mirrors how people naturally think through conditional logic.
That constraint is deliberate: a true/false structure means every transaction has a defined path. Nothing falls through the cracks because someone forgot an edge case. Whether the strategy is simple or a deeply branched tree, the logic remains readable, auditable, and safe.
And most importantly, it can be used and understood by any and all humans.
Testing, learning, iterating
We also built routing strategies as a first-class concept. Teams can create multiple strategies, enable or disable them, and compare performance over time. Routing should be continuously optimised and iterative, rather than set-and-forget. This gives payment teams the tools to treat routing as an ongoing optimisation practice rather than a one-time configuration.
AI is a composable node, not a black box
When we introduced AI-powered routing, we applied the same design philosophy. Rather than replacing the decision tree, AI sits within the legible flowchart as a node. A team can route all traffic through an AI node if they choose, or use it to optimise a specific branch while keeping the rest under manual control.
Our approach lets teams adopt it incrementally, within a framework they already understand, without sacrificing visibility over what’s happening and why. This matters because AI is one of many tools within a toolbox, and our trust in it isn’t binary.
The principle behind the product
The best infrastructure meets users where they are and helps them optimise their workflows on their own terms, rather than expect them to think like engineers overnight. We believe routing should be powerful enough to handle any level of complexity, and clear enough that the person configuring it can understand it immediately, regardless of their background or level of expertise.
Written by: Sophia Pistofidou - Product Lead
