Mobile Menu Button

Back to Spotlight

AI in Retail Finance: Why “Safe” Chatbots Fail, and What Comes Next

Insight
|
14 Jul 2026

Your compliance team didn't kill your AI strategy. Your chatbot did.


Most banks we talk to are caught in the same squeeze. The board wants AI in the product. The regulator wants accountability. So you ship a “safe” chatbot and call it progress.

Millions go into compliance guardrails, and what comes back is an AI legally allowed to answer FAQs and block a card. The customer still struggles. The board mandate is still unfinished.

The numbers say this is the norm, not the exception. EY-Parthenon's data points one way: only 16% of GenAI use cases in banking reach production, which means 84% never do. The barrier usually isn't the regulation. It's the shape of the solution.

So we explored a different approach: Scaffold Intelligence.
Scaffold Intelligence is an adaptive layer that sits on top of the app you already have. You don't throw the app away and you don't hand the screen to a chatbot. You scaffold it, the way scaffolding wraps a building you want to improve without tearing it down. The layer appears only where it helps, in the shape each moment needs, and steps back when it isn't needed.

Whatever shape it takes, the scaffold always does three things.

It contextualizes the past, instead of just showing it. It catches the free trial before it bills. It reveals the exposure you didn't know you had, like a stock position that quietly runs deeper than it looks once ETFs are counted.

It adapts to the moment, instead of staying static. It explains the drop behind a scary chart before you panic-sell. It turns a raise into an extra bit of savings, at the moment the money lands.

It anticipates the need, instead of waiting to be asked. It sets your savings target and plan from a single word, so you finish a task instead of starting a form.

And the part that makes it shippable: when the AI isn't confident, the scaffold quietly steps back and the customer gets the standard experience, uninterrupted. Nothing breaks.

Because the AI only prepares the path and the human makes the final call, meaningful human oversight is built into the architecture, not bolted on after. That's exactly the direction the EU AI Act pushes.

So the question worth asking your team this week: is regulation actually holding you back, or is it the architecture you chose?

If this shifted how you see it, let's connect on LinkedIn or Instagram, or reach out directly to continue the conversation.


Written by Soumava Mandal,
Principal Designer at zigzag
View all slides or download the PDF

Older

Event
Event

FIBE Berlin 2026 Recap: What’s Next for Retail Finance?

22 Apr 2026

Connect

Partner with zigzag

We are eager to hear more about exciting challenges and amazing opportunities to develop the latest products, services, and experiences. Don't hesitate to get in touch with us!

Contact for Business

Join Our Team

We're inviting you to become part of our dynamic team! It's a space for continuous learning, personal growth and meaningful collaboration. Don't miss out on the chance to join us!

Check Open Positions