E-commerce / Custom Manufacturing
Every shipping quote took half an hour. Dozens of times a day.
A custom AI-powered operations layer that closed the gaps the e-commerce platform couldn't, supporting a custom cabinetry business built on high-value orders.
Broken
Half an hour to build one shipping quote — dozens of times a day, often wrong.
Built
An operations layer over the store and the CRM: instant freight, one customer screen, OCR on invoices.
Now
25 minutes returned per quote, and 18+ inbound routes resolve without escalation.
01 // Before and after
Elapsed · 00:00:41
Elapsed · 00:24:18
· Measure each carton by hand
· Guess pallet configuration
· Look up freight class
· Call the carrier
· Re-quote — dimensions were off
The same quote, two different days of work.
02 // The challenge
The challenge.
Bridge the gaps the platform left open — accuracy, speed, and customer service capacity on high-value custom orders.
- 01Shipping quotes for bulky custom orders took 20–30 minutes each and were often inaccurate
- 02The platform couldn't swap colors on a configured build without manually re-adding every SKU
- 03Reps had to dig across disconnected CRM objects to get a full picture of one customer
- 04Manual invoice review before manufacturer handoff was error-prone and expensive when missed
Why off-the-shelf wouldn't work
This is a high-AOV custom-build business with a customer service team running parallel to a designer team, on top of an e-commerce platform and a CRM that were never built to meet in the middle. No off-the-shelf tool sits in that gap.
03 // What we built
01
Instant shipping quotes
Carton dimensions go in, a pallet-packing algorithm works out the configuration, and a freight estimate comes back before the customer has finished the sentence.
Order
Pallets
2 · 48×40×62
Weight
1,240 lb
Freight class
125
Estimate
$684.00
02
Cart automation
A finish change used to mean removing and re-adding every SKU on a configured build. Now one swap remaps the whole cart without touching a line item.
03
One-screen customer view
A phone number, email, or order ID assembles orders, chat history, designer notes, and invoices into a single panel, pulled in parallel from systems that never spoke.
Orders
3 open · 11 lifetime
Designer notes
Slate finish approved 04/12
Chats
2 threads · last 6 days ago
Invoices
1 pending manufacturer PO
One screen replaced six tabs and a guess.
Live · production
Inbound resolving without escalation.
intake routes · live
Live04 // Impact
Annualized impact
0
Minutes returned per shipping quote, dozens of times a day
Time to build a shipping estimate
Repeated dozens of times daily
05 // Technical implementation
Architecture overview.
A custom React tooling layer wrapping the e-commerce platform and CRM through their APIs, with parallel API orchestration behind the unified customer view and AI/OCR behind invoice review. A pallet-packing algorithm produces freight estimates in seconds. The intake system and chat run on the same data layer, so customer-facing and internal AI share one source of truth.
technical-highlights.md
- → Pallet packing algorithm for instant freight estimation
- → Platform API orchestration for live cart manipulation
- → Parallel API requests across CRM objects for the unified view
- → OCR-based invoice comparison against purchase orders
- → Intake system with 18+ conversational routes
- → After-hours chat triage into the same records
06 // Field notes
The platform wasn't wrong. It just stopped where the business kept going.
One screen replaced six tabs and a guess.
07 // Related case studies
Manufacturing / Enterprise
The Manufacturer's Operating System.
Seven connected systems on one data foundation, delivered in sequence — each live before the next begins.
Read case study
Automotive Services
The whole business ran on spreadsheets.
Six locations, each with its own spreadsheet and a tab for every day — replaced by one scheduling system, with an intake system covering the phones around the clock.
Read case study
Field Services / Multi-Location Franchise
The phone rings at 7pm. Nobody's there.
Nights, weekends, and storm-season surges — an AI voice system that answers, qualifies, and books the appointments the business used to sleep through.
Read case study
Let's talk about what you're trying to fix.
Thirty minutes. No pitch deck.