Archway

Inspiration

The gap in tech is not just about who can code, but who gets to design systems. As AI tools become easier to use, the advantage shifts to those who can architect workflows, data pipelines, and AI systems.
Existing system design tools are either too academic, too intimidating, or assume prior engineering thinking.

We observed a gap: many young adults, especially women, can use AI tools but struggle to build systems with them. Current tools either provide no guidance or overly generic tutorials. Neither supports real learning.

Archway bridges this gap with a guided playground that starts from your idea and helps you build a real architecture while preserving curiosity.

What It Does

Archway is a system design playground for beginners learning to think architecturally.

Users answer three questions about their idea, users, and workflows. Archway then generates a live architecture diagram with components like frontends, APIs, backends, and databases.

Learning happens through three lenses:

Product Manager: prioritize features and map user flows
Engineer: think about scale, databases, and failure points
Ethical Advisor: audit bias and inclusion with a live bias score

Additional features:

Deconstruction: reverse-engineer real apps through guided questions
Community: publish, fork, and collaborate on architectures like GitHub
Mentorship: experienced builders guide learners
Gamification: points and rewards, with incentives for inclusive design

How We Built It

Frontend: Next.js with a grid-based design system
LLM pipeline: generates initial architecture from onboarding inputs
Node-based canvas: interactive components with AI explanations
Persistent AI mentor: scoped to each user’s idea
Dynamic lessons: role-based exercises tied to the user’s architecture

Challenges

The main challenge was avoiding generic AI outputs. We refined prompting to ensure all content stays specific to the user’s idea.

We also reduced technical complexity to balance accessibility for beginners with meaningful depth for intermediate users.

Accomplishments

Bias Score: surfaces fairness issues at the architecture level in real time
Three-Lens Learning: mirrors real-world collaboration across product, engineering, and ethics

What We Learned

Beginners struggle with system design due to lack of clear starting points. Generating a first draft architecture makes learning easier by enabling critique and iteration.

We also learned that incentives shape behavior. Embedding inclusive design into rewards drives better engagement than separating it into standalone modules.