Neural Networks as a Controllable System: Design Functions

How the Kokoc Group team built a scalable operational AI approach for day-to-day design tasks in just one year, without compromising quality or consistency.

client

Kokoc Group

https://kokocgroup.ru/

Industry

Creative Agency, Performance, FinTech, Talent, Affiliate

Team size

12
design teams in 7 divisions

Example of use

SMM, branding, performance, landing pages, marketplaces

Results

Creative Production

80%

reducing the time spent on routine design tasks

From Hypothesis to Launch

by 62%

Faster campaign launches and landing page design

Variability

300%

an increase in the variety of creative assets and the frequency of testing

Subscriptions to Tools

12→1

One Phygital+ subscription instead of 12 separate subscriptions

Challenge

Kokoc Group is a multi-format group of companies that brings together its own and partner solutions into a single interface across the following areas: business performance, affiliate marketing, fintech, marketplaces, content marketing, and talent acquisition. Design teams work across different business areas and tackle challenges with varying KPIs, paces, and creative requirements. The goal was to build a unified AI-powered workflow system for all teams: one that avoids tool-related chaos while ensuring clear processes, predictable timelines, and consistent creative quality.

"We deliberately avoided building node trees that were too complex. Experience has shown that a simple working template scales better than a complex structure, especially when it comes to sharing an approach across teams."

— Daria Trubarova, Creative Director at Kokoc Group

Different business models, speeds, and KPIs

Kokoc Group is a unified, multi-format group of companies and a platform. The team had to tackle a wide range of tasks simultaneously—from in-depth brand development to rapid performance-driven creative work. The main challenge was to create a unified approach that took into account the different paces, requirements, and KPIs of each of the brand’s areas of focus.

Chaotic subscriptions and an overloaded process

Before the systematic implementation of AI, teams had disparate subscriptions to various neural networks and relied on manual processes in production. This complicated management, increased costs, and slowed down scaling: more coordination was required, there was duplication of tools, and quality varied across teams.

Key Objective of the Implementation

Move neural networks from an experimental phase to a managed operational system with the following design features: implement a unified stack, build clear pipelines, accelerate content delivery, maintain brand consistency, and achieve measurable business results in terms of timing and volume.

Decision

One platform.
Endless possibilities.

In 2025, the Kokoc Group team conducted an audit of tasks and tools, transitioned to a unified AI tools environment in Phygital+, identified scenarios with the greatest business impact, and assembled operational pipelines for routine production tasks. At the core are simple and scalable templates that can be shared across teams.

Midjourney

Nano Banana

ElevenLabs

VEO/Kling

"In branding, AI is particularly useful during the conceptual exploration phase: it allows you to quickly visualize several directions, compare options, and make a more informed decision before moving on to costly production."

— Daria Trubarova, Creative Director at Kokoc Group

Result

Over the course of a year of systematically implementing neural networks using Phygital+, the Kokoc Group team was able to accelerate the production process, shorten the path from concept to launch, significantly increase the variety of creative options, and create a unified, manageable suite of tools for all teams.

1. SMM: Across four accounts, the team produces up to 20 creative assets per month per channel; the time required for each creative asset has been reduced from 5 hours to 1 hour, and the number of variations has increased to 3–5.

2. Performance Funnel: up to 6 landing pages per month and up to 5 banner variations; the landing page development cycle has been reduced from 8 days to 3 days.

3. Branding: AI has accelerated the conceptual development process, testing of brand identities in digital and merchandise, and the preparation of visual directions prior to costly production.

4. Marketplaces: The generation of environments and semi-automatic adaptation have accelerated the preparation of product cards and the scaling of resizes.

5. Case Study: Kokos Soccer Club. Rapid development of visual concepts, exploration of several stylistic directions, and instant prototyping. AI-generated content and AI models replace expensive photo shoots, which significantly speeds up processes and saves money without compromising quality. Include prompts for AI-generated photo styles in the company's brand book to ensure consistency.

Watch Daria Trubarova's presentation on the implementation of neural networks at Kokoc Group

"Over the course of a year, AI has become an integral part of the operational design function: the time required to produce routine creative assets has been reduced by 80%, the time from hypothesis to launch has been cut by 62%, and the range of variations has increased by 300%, and a single Phygital+ subscription has replaced 12 disparate tools.”

— Daria Trubarova, Creative Director at Kokoc Group

Ready to turn AI into a working system for your team?

We can help you conduct a task audit, build operational pipelines, and integrate neural networks into your design and marketing processes to deliver measurable business results.