The Rise of the 'Brand Engineer'

Tommy Geoco observes a significant shift in the design industry where the traditional role of a designer is being subsumed by a new archetype: the 'brand engineer.' This shift is driven by the necessity to build internal tools that automate repetitive tasks, such as generating marketing assets or maintaining design system consistency. Geoco notes that 59% of designers are now building their own custom tools, moving beyond standard design software to create bespoke workflows that integrate directly with their product's codebase.

The Shift to Probabilistic Design Workflows

Geoco and the host discuss the transition from manual, deterministic design to a probabilistic workflow where the designer acts as a creative director. Instead of starting from a blank canvas, designers now use AI to generate dozens of variations, curating the best outputs and refining them in code.

  • The Role of Friction: Both speakers emphasize that total automation is not the goal. They advocate for 'preserved friction'—maintaining human oversight to fact-check intent and steer the AI.
  • Driving vs. One-Shotting: The speakers criticize the 'one-shot' mentality of evaluating AI models. They argue that true AI fluency is developed by 'driving' the model—iteratively refining its output through feedback and context-building until it aligns with a specific design language.

Building Context-Rich AI Systems

Success in modern design workflows depends on feeding AI models deep, domain-specific context. Geoco describes his process of creating 'context layers'—markdown files and rule sets that teach his AI how to handle specific components, such as using inner shadows instead of borders to simulate depth. This 'briefcase of context' allows the AI to produce outputs that are consistently on-brand without requiring constant manual adjustment.

The Future of Internal Tooling

Rather than relying solely on general-purpose AI tools, top design teams (like those at Ramp or Vercel) are building proprietary internal tools. These tools are tailored to the company's specific data sources and design constraints. The speakers predict that in the future, designers will carry their 'briefcase of context' across different tools, allowing them to maintain a consistent design brain that understands business logic, team feedback, and historical design decisions.

Key Takeaways

  • Embrace 'Vibe Coding': Don't be afraid to roll your own tools. If a task is repetitive, build a small utility to automate it.
  • Focus on Context, Not Just Prompts: Spend time building a 'context layer' (documentation, style guides, and feedback logs) that you can feed into your AI models.
  • Adopt the Creative Director Mindset: Use AI to explore a wide spectrum of possibilities quickly, then apply your critical thinking to curate and refine the best results.
  • Preserve Human Friction: Avoid fully automated, 'run-all-night' workflows. You need to remain in the loop to steer the model and ensure the output matches your intent.
  • Evaluate by Driving: Stop judging models based on single-prompt outputs. Evaluate them based on how well they respond to your steering and how quickly they learn your specific design language.