The future of AI compliance and model risk management, Yields, 2024
The scenario
Everyone is racing to put AI to work, but the truth is that models still “get things wrong” - quietly, and at a cost that only shows up much later - and compliance is now the price of running them at all. Yields is an award-winning SaaS platform that helps financial institutions and corporates stay in control of their models and AI systems. A platform that keeps both in check: traditional models and modern AI, centralised in one place, risk-scored, with responsibilities tracked and compliance provable at scale. Founded in Ghent in 2017, Yields had spent eight years quietly becoming one of the most awarded platforms in model risk. A tool for quants, turning into infrastructure for everyone. But the brand still spoke the language it was born in: technical, cautious, indistinguishable from every compliance vendor in the category. The product was already leading, however the brand was still asking permission - we, DPDK, were ready to change that.
The approach
DPDK was brought in for the full brand - identity, patterns, website. The identity earns its distinctiveness from the mechanism it exists to stop, and was built around a single creative decision: containment. The logomark is a butterfly, drawn as one unbroken outline. The butterfly effect: a small unseen error, a rounding, a drift - and a consequence nobody sized, arriving much later and somewhere else. That is what an unwatched model does. The mark is held inside a box, and the box is the argument. It is the exact point where the chain reaction is stopped - the effect, blocked at its origin, before it can become anything. Every surface obeys it. The patterns are that same mark, generated again and again. Rotated, scaled, layered, multiplied, take the butterfly, rotate it, repeat it - and shapes appear that were never drawn. Nothing is added: one outline, one small change, and the system returns something nobody designed. A brand language that behaves exactly like the thing it governs, where a minor misreading produces a result you didn't predict.
Inspired by Refik Anadol, the uniqueness
Yields' imagery had to be made of the material the company is built on. Not people at laptops, not dashboards, not abstraction: data. A single, unexpected reference: Refik Anadol. The artist who takes vast archives - 113 million images of New York, 200 million photographs of the Earth, a century of an orchestra's recordings - and feeds them to a machine until they come back as image matter. Fluid, volumetric, alive. His whole subject is the unseen world of data: making visible what the eye was never able to see. That is Yields' business: institutions are running on data they cannot see - thousands of models, millions of decisions, a landscape nobody has ever actually looked at. The blobs are not decoration but data, given a body: dense, volumetric, in flux, generated with AI and art directed through abstract as something you can see. Where Anadol tunes toward the sublime, we tuned toward the governed.
The impact
Yields stopped looking like compliance software and the brand finally matches the success of its product. Yields walks into a room full of banks and regulators no longer looking like one more compliance vendor, but like the company that defines the category it already leads. A brand that speaks in a language entirely its own - built from its own subject matter. The blobs carry the scale of what Yields governs, the interface stays plain where the proof lives, and the identity holds the two together instead of decorating either. It's a brand that doesn't talk about control - it's built out of it. Every surface, every pattern, every image - grown from data, and never out of hand.