Wever Callback

Inspiration

Consignment boutiques depend on people bringing in the right inventory. But the same small team answering questions about clothing and handbags is also serving customers, evaluating items, and running the shop.

An interested consignor may need a conversation before arranging a visit. When follow-up gets delayed, both the customer and the shop are left waiting.

I built Wever Callback to help boutiques move from an incoming inquiry to a prepared consultation, with the conversation and next steps organized in one place.

What it does

Wever Callback is a customer follow-up application for consignment boutiques.

A customer submits a callback request describing what they would like to consign. The business reviews the inquiry and approves an AI-assisted phone interview through CALL-E. The interview gathers item details, condition, questions, and a preferred consultation time.

The business can review the saved conversation and outcome, add follow-up notes, and record a booking after confirming the arrangement with the customer.

The application includes:

• A customer callback request form.
• An inquiry inbox with separate live and sample workspaces.
• Business setup for approved facts, interview questions, and boundaries.
• Individually approved CALL-E interviews.
• Saved conversations and structured outcomes.
• Follow-up notes and manually confirmed appointment records.
• Sample scenarios covering an interested consignor, a pricing question, and an opt-out.

Questions beyond the business’s approved information are referred to a person. A requested consultation time remains unconfirmed until the business agrees on the arrangement with the customer. The application records that confirmation; it does not create calendar events.

How I built it

The application uses TypeScript, React, Vinext, and Tailwind CSS, with Cloudflare Workers and D1 for the hosted application and persistent records. Drizzle ORM manages the database schema.

The live interview integration uses CALL-E’s Calls API. It sends the business’s interview instructions and a requested result structure, then retrieves the conversation and outcome for review.

The community contribution includes a runnable local version with SQLite-backed records and sample rehearsals. The sample workflow requires no calling credentials or credits. Live calling requires the operator’s own CALL-E account and explicit approval.

Challenges and what I learned

The hardest part was making the next step clear when a phone call produces an incomplete or uncertain result. A failed connection should remain a failed connection, and a preferred appointment time should remain a request until someone confirms it.

I completed a real 110-second CALL-E interview on September 10, 2026. Later live attempts on September 12 failed to connect and were reported to CALL-E support for investigation.

The submitted video demonstrates a fictional customer and a simulated conversation. It shows the saved inquiry, conversation, consultation request, and manually recorded booking without exposing the private test call.

The local contribution passed 30 mocked application checks, TypeScript checking, and a production build. These checks validate the application separately from the live phone test.

I learned that useful automation must leave a busy shop owner with enough context to act: what the customer has, what they need, what remains unanswered, and what should happen next.

What’s next

The next step is a pilot with independent consignment boutiques to test whether this workflow fits their daily operations.

I would measure response time, staff time spent on follow-up, and the proportion of inquiries that reach a confirmed consultation. These are goals for validation, not results already achieved.

Future development would focus on calendar integration, shop-specific intake requirements, and improvements informed by actual boutique use.

Built With

• call-e
• cloudflare-d1
• cloudflare-workers
• drizzle-orm
• react
• tailwindcss
• typescript
• vinext