AI Keynote: Navigating the New Era of Creative AI Workflows
This text, generated using AI, accompanies the full webinar available on our Vimeo channel.
May 13 | 11:00—12:00 | Keynote
Creative AI Workflows – Maintaining creative sovereignty from brainstorming to the final shot
The world of digital creation is moving at a breakneck pace. For those of us in the creative industries, it’s no longer a question of if we should use AI, but how we use it without losing our unique artistic voice. In a recent lecture, Lenka Hámošová, a Prague-based designer and researcher who has been “tinkering” with AI since 2018, shared a roadmap for maintaining creative agency in an age of automated art.
The Evolution: From Messy Hacking to AI Agents
Hámošová frames the current state of AI as a natural step in the historical evolution of animation tools, moving from the physical zoetrope to computer-generated 3D, and now to generative AI.
We have transitioned through three distinct eras:
- The Messy Hacking Period (2018–2022): A time of niche experimentation using tools like Google Colab, where artists “broke” models to see what was inside.
- The Mass Accessibility Breakthrough (2023): The year AI became widely known to the public through models like DALL-E and Sora.
- The Workflows and Agents Era (2024–2026): Our current reality, where we no longer just use one tool; we build pipelines.
Today, concepts like “vibe coding” allow creators to generate entire apps or marketing strategies in an hour, while AI agents automate complex multi-step processes.
Workflow Thinking: Connecting the Nodes
The most significant shift in professional AI practice is Workflow Thinking. Instead of throwing a prompt at a machine and hoping for a good result, creators are building custom “pipelines” inspired by visual coding.
A primary tool in this space is ComfyUI, a node-based editor.
- What it does: It allows you to connect different AI models (like Stable Diffusion or AnimateDiff) as “nodes” on a canvas.
- The Benefit: You can repeat complex processes over and over by simply changing one variable, giving you much higher artistic control.
💡 Key Insight: The “Egg Metaphor” In the 1950s, cake mix manufacturers realized that housewives didn’t feel they were “baking” if they only added water. Once they removed the dried eggs and required the user to add a fresh egg, sales skyrocketed because it restored a sense of agency. As a creator, don’t let AI do everything—ensure you are “adding your own egg” to the process to maintain your authorship.
Case Study: High-Level Creative Control
How does this look in practice? Hámošová highlights several groundbreaking projects:
- Real-time Collaboration: At a TED Talk in Prague, a team used Stream Diffusion to generate visuals driven by a live dancer. Three computers processed the music and text prompts in real-time to mask the AI generation directly onto the dancer’s moving body.
- The Paul Trillo Workflow: For a music video for the artist Kuko, Trillo’s team didn’t just “generate” a video. They trained a custom LoRa model on 60 original hand-drawn illustrations to maintain a consistent style, then moved from 2D assets to 3D extruded environments, and back to 2D using “Generative Ink and Paint”.
Tomáš Rampula’s Artistic Mastery: Czech artist Tomáš Rampula spends months on single projects, manually retouching “master frames” in Photoshop before using AI to animate the sequences. This “layered” approach ensures the AI is an extension of his hand, not a replacement for it.
Protecting the Human Element
As AI becomes more photorealistic, Hámošová argues that our embodied human experience becomes our most valuable asset. AI can cognitively understand emotions, but it cannot feel them because it lacks a body.
To keep your work unique, you must identify your “Unique Toolkit”:
- Your Craft: The skills you’ve spent years mastering.
- Your Obsessions: The weird, niche interests that only you have.
- Your Lived Experience: The suffering, joy, and sensory memories that AI cannot replicate.
⚠️ Warning: The “AI Slope” Many AI tools have “explore feeds” that showcase popular generations. While good for inspiration, they can lead to a “blind adaptation” where everyone repeats the same aesthetic patterns, resulting in content that feels “empty” or lacks depth.
Next Steps: Map Your Process
To take control of your creative future, Hámošová suggests a practical exercise: Reverse engineer your workflow.
- Mind Map your typical project: Break it down into the smallest possible boxes—from where your ideas are born (like the shower!) to the final export.
- Color-code the boxes: * Keep for Yourself: Tasks you love or that require your unique human touch.
- Automate: Repetitive, boring, or non-creative tasks that AI can handle.
Collaborate: Areas where you and the AI can iterate together.
