Jeremy Blaze, founder of Never Before Seen (NBS), has developed an internal AI-powered design library that helps increase his team’s leverage by providing quick access to the agency’s decade of past work. The tool, which Blaze began building in late 2024, serves as a searchable repository of institutional knowledge gathered over 10 years that allows his team to find relevant design references in seconds rather than hours. His approach demonstrates how effectively organizing past work can transform into a powerful resource for specialized design teams.
Under Blaze’s leadership, Never Before Seen operates as an agency that partners closely with early-stage startups to execute on product and design. The firm has built its reputation on solving intricate design problems with this internal tool reflecting not just efficiency but how AI can level up the way we do work.
From Experience to Implementation
Blaze’s idea for his agency’s design library emerged from inefficiencies he observed in traditional design research processes. While designers regularly look to external sources like Mobbin, Dribbble and more recently, v0 and Lovable for inspiration, Blaze recognized that their most valuable reference point was often their own extensive portfolio developed over 10 years of designing products.
The challenge wasn’t a lack of past solutions but the difficulty in accessing them efficiently. Blaze could locate specific files from memory, but new team members faced a significant hurdle in navigating this vast collection of work. As he put it, finding the right files was “fine for me, given I know all the projects. But for anyone joining, I wanted them to be able to access company knowledge from before their tenure.”
The system was developed through a two-phase approach. Phase 1 involved creating a comprehensive, searchable design library— essentially a custom version of Mobbin, the popular public reference site for software design, tailored specifically to Never Before Seen’s work. This initial foundation compiled all screens in one centralized location and then used GPT Vision to automatically tag them based on functionality, layout and design style. This enables team members to find specific elements like “dark mode dashboard” instantly. But this only solved part of the problem.
Phase 2 enhanced this by implementing AI-powered discovery capabilities that recognize patterns in current design challenges and find solutions in past works. This advancement allowed team members to describe projects in natural language— for example, “Help me design a transaction list for a fintech app”— and receive relevant screens even without explicit matching tags.
Simplifying Internal Design Processes
One of the most significant benefits of Blaze’s internal tool in Never Before Seen is how it The true strength of Never Before Seen’s internal design library lies in how it surfaces the agency’s deep well of experience for the entire team, regardless of tenure. By transforming a decade of product design into an instantly searchable, AI-augmented resource, the tool enables anyone at the agency to access solutions and insights from across its history. Whether referencing a niche workflow or rapidly prototyping during client sessions, team members can leverage proven patterns and hard-won knowledge in seconds, ensuring no valuable lesson is lost to time.
This approach does more than streamline processes—it fundamentally elevates the agency’s creative output. With the collective wisdom of Never Before Seen always within reach, the team is empowered to solve complex challenges with greater confidence and originality. As the library continues to evolve, it stands as a testament to how thoughtfully applied AI can transform expertise into a living asset, raising the bar for every project and shaping the future of design work.
Blaze’s AI-powered internal design library at Never Before Seen transforms a decade of expertise into actionable insights. What started as a time-consuming and difficult process of finding past work has become a system that, as Blaze confirms, “turned hours of searching into seconds.”