Building AYA with FlukeBase – A Technical Deep Dive

AYA is not just a product – it is also a case study in modern full-stack development. This article walks through the technology choices behind AYA and how FlukeBase accelerates the development workflow.

The Tech Stack

AYA is built on a modern JavaScript/TypeScript stack optimized for rapid iteration and production reliability:

• Next.js 15 with the App Router provides server-side rendering, API routes, and static generation in a single framework. The /app directory structure keeps routes, layouts, and server components organized.
• Prisma ORM handles database access with type-safe queries and automatic migration management. The schema serves as both documentation and runtime contract.
• PostgreSQL stores user data, projects, chat histories, component databases, and pricing records.
• Claude AI (via the Anthropic SDK) powers the requirements generation, component analysis, and conversational interface.
• Zustand manages client-side state with minimal boilerplate.
• Tailwind CSS v4 with a custom design system provides consistent styling across all components.

API Architecture

AYA uses Next.js API routes for all backend logic. Key endpoints include:

/api/chat-stream - Streaming AI conversation /api/requirements - Hardware requirements generation /api/pricing - Component pricing queries /api/articles - Blog content management

Each route follows a consistent pattern: authenticate the request, validate input with Zod schemas, execute business logic, and return structured JSON responses.

FlukeBase Integration

FlukeBase serves as the deployment and operations platform for AYA. The integration covers several areas:

Deployment Pipeline

The CI/CD pipeline is straightforward:

1. Push code to the main branch on GitHub.
2. GitHub Actions runs on a self-hosted runner: lint, type-check, build, and test.
3. On success, the Docker image is built and deployed to the VPS via CapRover.
4. Health checks confirm the deployment is live.

The entire cycle from push to production takes under five minutes.

MCP Tools

FlukeBase provides Model Context Protocol (MCP) tools that integrate directly with the development environment. These tools enable:

• Project management: Create tasks, track progress, and coordinate work across features.
• Deployment control: Trigger deploys, check status, and roll back from the CLI.
• Monitoring: View application health, error logs, and performance metrics.
• Memory: Persistent context that survives across development sessions.

Self-Hosted Git

In addition to GitHub, AYA mirrors its repository to FlukeBase’s self-hosted git platform. This provides redundancy and enables tighter integration with FlukeBase’s CI pipelines.

Development Workflow

A typical feature development cycle in AYA follows spec-driven development:

1. Requirements: Define what the feature should do, with acceptance criteria.
2. Design: Document the technical approach, data models, and API contracts.
3. Tasks: Break the implementation into discrete, testable units.
4. Implementation: Build against the spec, with automated checks at each step.
5. Validation: Verify the feature works in production, not just locally.

This structured approach prevents scope creep and ensures every feature ships with documentation and tests.

Contributing

AYA is under active development. If you are interested in AI-assisted hardware design tooling, the codebase is accessible and well-documented. The combination of Next.js, Prisma, and Claude AI makes it straightforward to add new capabilities – from additional supplier integrations to new analysis modes.

Check out the repository and start a conversation. We welcome contributions from hardware engineers, software developers, and anyone interested in bridging the gap between design intent and physical implementation.