Carbon for AI is an extension of the Carbon Design System, designed to give AI instances in IBM products a visually and behaviorally distinct identity. As AI integration grows, transparent AI is essential to maintain user trust. This pattern asset library and accompanying guidance is critical to IBM's AI mission and business strategy.
The Carbon for AI team was tasked with developing AI guidance for IBM Software, focusing on labeling, transparency, ethics, and explainability. Due to my success leading IBM’s Figma implementation strategy, including work on the Carbon Design System, I was selected to lead Figma strategy for this initiative. Initially brought on for a two-week loan, my role expanded into an ongoing partnership with IBM CIO and IBM Software, given the mutual benefit of the collaboration. The effort is led by Adam Cutler, IBM Distinguished Designer, and Ramiro Galan, Senior Designer.
To meet this challenge, the Carbon team embedded experienced designers into AI working groups to ensure alignment with Carbon principles while addressing AI-specific design needs. I was tasked with leading the created an AI Essentials Figma kit to provide design and development teams with a consistent, scalable framework for integrating AI elements. The team worked in agile sprints to enable rapid feedback and iteration, ensuring the AI components integrated seamlessly into the broader Carbon library. My approach consisted of 3 parts — understand, build, test — I first familiarized myself with the existing component and pattern explorations, read the guidelines and best practices,
One of the key challenges of this initiative was managing version control for Phase 1 and Phase 2 components, which often led to confusion between designers and developers regarding which assets were ready for production versus still in development. To mitigate this, I implemented a clear versioning system within Figma, organizing libraries by phase and enforcing strict naming conventions to ensure alignment across teams. Another challenge involved preparing design files for handoff to developers, as overlapping work between phases frequently caused misunderstandings. To address this, I streamlined the hand-off process by clearly labeling and annotating files to distinguish between phases, reducing confusion. Communicating changes and updates to design specs also proved difficult, so I established a centralized system for real-time updates and detailed change logs to keep all stakeholders informed. Additionally, Figma's limitations, such as the lack of native support for gradients, variables, and advanced masking at the time, required me to develop new techniques that worked around these constraints, enabling the team to maintain high-quality designs. These combined efforts ensured smoother collaboration and a successful rollout of the Carbon for AI project.