Customer to crew GPT

Inspiration

We noticed that SMB like HVAC and plumbing SMBs juggle constant customer calls, scheduling chaos, and technical documentation spread across binders and spreadsheets. These businesses need affordable AI tools that make their lives easier, without heavy integrations. That’s where our idea came from: build two GPTs that work together — one for customers and one for employees.

What it does

Customer GPT: Acts as a concierge. It triages issues in plain English, checks for hazards, gives ballpark price ranges, and books appointments.

Employee GPT: Acts as an internal copilot. It reads company documents (Appointments, Device Knowledge, Inventory, Pricing) to prepare job packs, reserve parts, generate Good/Better/Best quotes, and confirm work orders.

Together, they bridge the gap from customer to crew — turning a vague “my AC is blowing hot air” into a scheduled visit, reserved parts, and a ready-to-go diagnostic checklist.

How we built it

Created system prompts for both GPTs.

Prepared four supporting documents: Appointments, Device Knowledge, Inventory, and Pricing.

Wired the prompts so the GPTs use those docs as their “source of truth.”

Added lightweight “Actions” that simulate confirmations (Appointment booked, Parts reserved) with reference IDs, no backend required.

Challenges we ran into

Designing prompts that feel natural for two different audiences (customers vs employees).

Keeping outputs structured (JSON/job packs) while still user-friendly.

Making “fake Actions” convincing without any external APIs.

Packing the workflow into a demo-friendly, hackathon‑sized scope.

Accomplishments that we're proud of

Built two GPTs that complement each other in a real SMB workflow.

Clear, realistic sample data docs that showcase scheduling, manuals, parts, and pricing.

What we learned

SMB workflows are complex but documents already hold the answers — GPTs just need to read and structure them.

Splitting responsibilities (customer-facing vs employee-facing) reduces prompt clutter and makes outputs sharper.

What's next for Customer to crew GPT

Add live connections to calendars, CRMs, or inventory systems.

Expand Device Knowledge with richer manuals and troubleshooting trees.

Train on real intake transcripts to improve customer triage.

Build a simple web front end to showcase both GPTs working together in real time.

Built With

• gpt