AI Media System
Context
Internal Media Team, Kognity
Role
Media Production Specialist
Focus
Creative systems
AI media
Prompt strategy
As Kognity shifted toward an AI first media production, the challenge wasn’t generating more images, it was preserving creative intent and brand coherence while redefining how visual work was produced.
I led the evolution of Kognity’s visual language for AI generated media, acting as the bridge between existing human design work and a new, scalable prompt driven approach that established what Kognity’s new “AI media” should look like going forward.
Defining the Framework
For AI generated media to operate at scale without compromising quality, the system required clearly defined visual standards, evaluative logic, and structural consistency, not just effective prompts.
I focused on translating Kognity’s existing design language into a framework that could guide both generation and assessment, ensuring outputs remained aligned with the intended visual direction.
AI Media Judge
To maintain consistency at scale, an evaluation layer was introduced to assess visual clarity, structural fidelity, and alignment with defined style rules.
— Defined scoring criteria for visual alignment and clarity
— Structured outputs into evaluative metadata fields
— Enabled retry logic for substandard generations
— Escalated unresolved outputs to human review
Prompt Library & Visual Style Framework
Alongside evaluation, a structured prompt framework encoded Kognity’s visual language into reusable, scalable components.
— Established modular prompt components (Container Prompts and Variable Prompts)
— Defined naming and versioning conventions
— Ensured cross subject visual consistency
— Documented standards for future contributors
LEGACY VISUAL LANGUAGE
updated VISUAL LANGUAGE
Outcome
The framework introduced a structured layer of creative control into AI generated media production, reducing visual drift and increasing consistency across subjects.
It enabled scalable experimentation across multiple image models while maintaining alignment with Kognity’s established visual standards.
Manual quality review was reduced, and a reusable foundation was established for future AI driven content workflows.
Reflection
This project clarified where I create the most value, not in producing isolated outputs, but in shaping the creative direction that guides them.
I’m driven by defining what work should look and feel like at scale, especially when emerging tools challenge established standards and demand a new visual language.