SILVERSIDE And Hyde Launch Atelierbench, Revealing AI Models Still Fall Short Of Professional Design Standards

(MENAFN- ACCESSWIRE) SAN FRANCISCO, CA / ACCESS Newswire / October 20, 2026 / AI can generate an image in seconds. But can it produce work a professional designer would actually ship to a client?

AI innovation company SILVERSIDE and Hyde, an AI start up which brings model sovereignty to enterprises, have built AtelierBench, a first-of-its-kind benchmark designed to evaluate AI against the real-world judgment and workflows of professional creatives.

While most AI benchmarks score a single image against a short prompt or crowd taste, SILVERSIDE and Hyde test AI against real client briefs, expert judgment-based scoring criteria, and the multi-step workflows designers actually use. The goal is to answer a question that will become increasingly important as AI takes on a larger role in creative production: can AI meet the professional bar, rather than simply generate creative output?

"AI has gotten incredibly good at making things, but making something and knowing whether it's good are two very different skills," said PJ Pereira, founder of SILVERSIDE. "Creative people make hundreds of decisions between receiving a brief and delivering the final work. We want to understand whether AI can learn to navigate those decisions too."

"Most AI evals are built around problems with a right answer," said Anirudh Sanga, co-founder and CEO of Hyde, "Creative work doesn't have one, and that's what makes it interesting. AtelierBench captures the judgment that separates good design work from great: can a model deliver an output, would a real design team actually ship it, and what did it cost to get there? That's a much more honest definition of 'great' at creative work, and once you can measure it, you can build models that are great at it."

AtelierBench shows professional design is far from saturated. Among the benchmark's key findings:

None of the models tested meet the professional bar. Weaker models often fail to execute design workflows at all, while even the strongest models miss quality details that matter to brands. As Atelier scores across granular quality dimensions, we see that models are weakest on precision, especially text and logos and staying true to reference images. A misspelled tagline, distorted logo, or off-brand change can make an output unusable, however good the rest is. Harness engineering improves performance to a degree, but after that gains plateau and costs keep rising. For instance, doubling the tool budget adds at most six percentage points of execution, and in some cases heavier setups cost more than 50% extra with no quality gain. Harnesses can actually make the strongest models worse.

For SILVERSIDE, which works with global brands to apply generative AI to commercial creative production, the benchmark addresses a practical challenge facing creative teams today. New models arrive constantly with technical scores and leaderboards, but those rankings reveal little about how well a model will perform against the realities of a professional creative brief.

For Hyde, the results reinforce the opportunities in continuous evals, model routing and post-training. Together, these can improve performance while giving enterprises more control over the models, data and IP that underpin their AI workflows.

"Ultimately, creatives don't need another score telling them how powerful a model is," Pereira added. "They need to know whether it can help them make better work."

About SILVERSIDE

SILVERSIDE is an AI innovation lab helping the world's leading brands revolutionize their marketing ecosystems. Founded in San Francisco, SILVERSIDE brings together creative technologists, strategists and builders to combine human creativity with cutting-edge AI. The company partners with brands including Coca-Cola, Amazon, Sephora, Panasonic, BMW and others to create ambitious AI-powered content, build scalable creative systems and transform how marketing organizations work. Learn more at silverside.

About Hyde

Hyde enables enterprises to transition from rented to sovereign AI. Through its training platform and forward-deployed approach, Hyde works with organizations to build private evaluations, and fine-tune and deploy specialized models using enterprise data and expertise. Models built using Hyde have outperformed the frontier on accuracy, latency and cost. Hyde is backed by Bain Capital Ventures and General Catalyst, with AI engineers from Palantir, Deepmind, and Optiver. Learn more at hyde.

Media Contact

Emma Swanson
Daughter Comms
[email protected]
+1 716-225-1186

MENAFN20102026004220003708ID1111835088