Thredd: Explainable AI for Intelligent Commerce Decisions

"The world is changing rapidly," says Edwin Poot, Chief Technology Officer at Thredd. 

Artificial intelligence may be relatively new in financial services, but it is already redefining how decisions are made in digital commerce. 

Instead of systems that simply process transactions, organisations are now building intelligent decision layers that interpret intent, understand context and act through APIs in real time. 

This shift turns payments and commerce flows into a continuous intelligence engine that helps clients make better, faster and more defensible decisions.

From transaction processing to intelligent processing

Traditional payment and commerce platforms were designed to maximise throughput and reliability. Their job was to take in a transaction, apply a set of static rules and output a yes or no.

As AI matures, that model is no longer enough. 

Intelligent processing adds a cognitive layer that can interpret the intent behind a transaction, consider contextual signals and draw on external data sources via APIs.

Rather than focusing purely on volume, this open intelligence approach helps organisations optimise risk, experience and revenue simultaneously. It supports decisions that are tailored to individual customers and situations, not just broad segments.

AI agents and the new trust equation

The rise of AI-powered or agent-led commerce introduces a fresh set of challenges. Businesses are increasingly asking software agents to make decisions on their behalf, from risk approvals to personalised offers.

For that to work, agents must have access to real-time APIs so they can "read the context, understand what's happening" and respond accordingly.

However, technical capability alone is not enough. Companies must be able to trust that the agent will consistently do the right work for them and their customers. 

That pushes issuers, processors and merchants to move away from static decision flows towards more intelligent ways of exposing and governing their services.

Why explainable AI is now non-negotiable

When a card or transaction is declined, customers want to understand the reason. 

As Edwin explains: "Just saying yes or no is not enough anymore". 

Explainable AI is, therefore, becoming a non-negotiable requirement. It ensures that organisations can clearly show why a transaction was approved, declined or flagged and on what basis.

This transparency strengthens trust with clients and consumers, but also with regulators and authorities. 

Decisions are no longer made solely on a fixed set of rules. Instead, they are based on constant adaptations, as models adjust to new behaviours, threats and use cases. 

Explainability provides the audit trail that turns those adaptive systems into something defensible and supervisable.

Continuous learning at the core infrastructure layer

The most powerful change lies in how models evolve. By looking at every transaction and learning from it, fraud, credit and personalisation models can update automatically instead of waiting for manual tuning cycles.

Each event becomes a data point that refines risk thresholds, improves detection accuracy and sharpens targeting.

"This makes our infrastructure more intelligent,” says Edwin.

Rather than bolting AI on at the edges, the intelligence is embedded in the core decisioning layer. That layer becomes the capability that allows institutions to adapt in real time while still explaining every decision. 

In an era of intelligent commerce, those who combine adaptive models with transparent reasoning will be best placed to earn and keep customer trust.

"Instead of focusing just on processing transactions, it is more of an open intelligence play, facilitating our clients to make better decisions."

"Explainable AI ensures that we can always show why a transaction was approved or declined, which builds trust with clients, consumers and authorities."

"Looking at every transaction and learning from it means fraud, credit and personalisation models evolve automatically over time, making our infrastructure more intelligent."