Agentic AI · Shipped Q3 2026
Inquiry Triage Pipeline
How abestack.com captures leads on a static Vercel site, queues them in Firestore, and runs Genkit triage before a human coordinator replies.
Firestore
Inquiry queue
Genkit
Auto triage
Human
Review gate
Fallback
FormSubmit
Problem
Static marketing sites usually stop at “email delivered.” That works until volume grows — then there is no queue, no urgency signal, and no structured way to spot agentic AI opportunities before the discovery call.
Solution
- submitInquiry — HTTPS endpoint writes raw form data to
inquiry_queue - onInquiryCreated — Firestore trigger runs a Genkit triage flow (summary, category, urgency, audit flag)
- FormSubmit fallback — if Firebase is unreachable, the form posts to email backup automatically
- Thank-you page — inline success UX without leaving the brand
Architecture
abestack.com (Vercel static)
→ POST submitInquiry (Cloud Function)
→ Firestore inquiry_queue/{id}
→ onInquiryCreated → Genkit triageInquiry flow
→ Coordinator reviews in Firebase Console
→ Human reply via hello@abestack.com
Results
Every inbound lead is structured before reply. Agentic AI audit requests are flagged automatically. The same pattern ships on client projects — this case study is the live reference implementation.
Technology
Cloud FunctionsFirestoreGenkitGemini 2.5 FlashFormSubmitVercel
"We dogfood the same agentic stack we sell — static site on Vercel, intelligence on Firebase."