Generated Moon

Problem

Design agencies lose hours to back-and-forth discovery calls before a project even starts. Generated Moon replaces that with an AI-led intake flow: the chatbot conducts a structured 5–8 question assessment, then auto-generates a detailed project brief — scoped, formatted, and ready for proposal. Stripe handles payment, Resend sends confirmation, and Calendly closes the booking loop.

Architecture

Key Features

• AI Assessment Flow: Chatbot conducts a structured 5–8 question intake — budget, timeline, design aesthetic, deliverables — and produces a formatted project brief automatically.
• Lead Capture System: Every assessment response is stored via /api/leads, giving the agency a complete CRM pipeline without third-party tools.
• Stripe Integration: Clients can pay deposits or retainers directly within the flow — no external invoicing step.
• Resend + Calendly: Automated confirmation emails and discovery call scheduling close the client acquisition loop end-to-end.
• Dual LLM Support: Abstracted provider layer supports Claude AI for premium quality and HuggingFace Mistral-7B as an open-source fallback.

Technical Challenges

• Conversational State Management:Maintained multi-turn assessment context across the chatbot session without leaking prior responses into brief generation.
• Structured Output from LLM:Engineered prompts to reliably produce consistently-formatted project briefs regardless of open-ended user input.
• Full-Stack Payment Flow: Integrated Stripe webhooks with Express session state to gate brief delivery behind confirmed payment.

Impact

End-to-end client acquisition platform that demonstrates applied LLM product engineering: conversational flow design, structured output extraction, payment integration, and full-stack delivery. Deployable on Railway or Vercel with zero infrastructure overhead.