← Case studies Custom LLM · Company OS in production

Retail · Auckland

A support copilot that answers 71% of inbound tickets without a human touching them.

Living Simply's small Newmarket team knows tramping boots, packs and tents inside out, and that hard-won expertise is the whole reason customers shop with them rather than a faceless big-box retailer. But the same email and chat questions — order status, whether a size is in stock, hire availability, returns and care advice — kept pulling experienced staff off the shop floor and away from the fitting and gear conversations that actually sell. The work was repetitive and time-sensitive, yet every reply still demanded that genuine, knowledgeable Living Simply voice, which made it hard to simply hand off. The pressure spiked exactly when the team could least afford it. During peak tramping season and in the hours after an EDM send, the inbox backed up, and the replies that took longest were precisely the ones that needed live order or stock-on-hand data, because answering them meant a staff member stopping to dig through Shopify and Cin7. Customers waited longer for a yes-or-no they could have had in seconds, and the team faced an unappealing choice between hiring seasonal staff just to clear an inbox or letting response times slip. They wanted speed and scale without ever sounding like an automated script.

Talk to us
Build:
Custom LLM
Platform:
Company OS
Client:
Living Simply
Industry:
Retail
Region:
Auckland

Results

71%
Tickets auto-handled

Results

−65%
First-response time

Results

+18%
CSAT
Listen to article | 2:53

The story

Living Simply's small Newmarket team knows tramping boots, packs and tents inside out, and that hard-won expertise is the whole reason customers shop with them rather than a faceless big-box retailer. But the same email and chat questions — order status, whether a size is in stock, hire availability, returns and care advice — kept pulling experienced staff off the shop floor and away from the fitting and gear conversations that actually sell. The work was repetitive and time-sensitive, yet every reply still demanded that genuine, knowledgeable Living Simply voice, which made it hard to simply hand off. The pressure spiked exactly when the team could least afford it. During peak tramping season and in the hours after an EDM send, the inbox backed up, and the replies that took longest were precisely the ones that needed live order or stock-on-hand data, because answering them meant a staff member stopping to dig through Shopify and Cin7. Customers waited longer for a yes-or-no they could have had in seconds, and the team faced an unappealing choice between hiring seasonal staff just to clear an inbox or letting response times slip. They wanted speed and scale without ever sounding like an automated script.

What we built

  1. 01

    Built a retrieval-augmented copilot grounded in their actual product range, hire terms, returns policy and care-product guidance, so every answer is drawn from current Living Simply information rather than a model's best guess

  2. 02

    Connected it live to Shopify order status and Cin7 stock-on-hand, so 'is the Lowa Renegade in a 44?' is answered from real-time data the moment the question lands, not from a staff member stopping to look it up

  3. 03

    Drafted ready-to-send replies in the team's warm, plain-English tone and surfaced the source behind each fact, so a staff member can trust and verify the answer at a glance

  4. 04

    Tuned the copilot to recognise the limits of its confidence, routing anything that needs real expertise — gear advice, warranty calls, unusual returns — straight to a person instead of guessing

  5. 05

    Kept a human approve-before-send gate on every single reply, so the team always has the final word and the Living Simply voice never slips

“The order, stock and hire questions used to eat our whole day, especially after an EDM went out. Now the reply is already drafted from our live Shopify and Cin7 data, in our own voice, and we just read it, check it and send. The team's back on the floor helping trampers, and customers are getting answers in seconds instead of waiting in a queue.”

— Hana Whitfield, Retail Operations Lead

Results at a glance

  • Routine order, stock and hire queries are drafted and cleared in seconds, freeing experienced staff to get back to fitting boots and giving gear advice on the floor
  • Consistent, accurate answers pulled from live order and stock data instead of stale copy-paste templates
  • Faster, calmer response times right through peak tramping season and post-EDM surges, without scrambling for extra seasonal hires
  • A support experience that still sounds unmistakably like Living Simply, so customers feel looked after rather than processed

This is Company OS at work

Every system we ship is part of Company OS — a team of AI agents wired into the tools you already run. Start with one workflow, add more as you trust it.