Case studies/Irrigation distributor
Irrigation distributor · Mbombela

From 4 hours to 12 minutes.

An irrigation parts distributor in Mbombela was losing half their week to quote admin. We connected their inbox, ingested 18 months of past quotes, and shipped a Telegram approval flow. The first week, they sent 87 quotes in the time they used to send 12.

IndustryIrrigation distribution
LocationMbombela, MP
Team size14 staff · 4 reps
PlanQuote Agent (Pro)
Time to live4 weeks
STAGE 01 · INBOX Greenfields Co. PURCHASING@DISTRIBUTOR.CO.ZA "Pricing on 24× RB-5000 sprinkler heads, 8 rolls 17mm drip line, 6× valves" PARSED · 3 SKUs FOUND STAGE 02 · APPROVAL @HorizonQuoteBot JUST NOW · TELEGRAM Draft ready for Greenfields: R 18,420.00 MARGIN 38% · 3 LINE ITEMS ✓ APPROVE & SEND STAGE 03 · DELIVERED QU-2026-0418.pdf SENT · BRANDED CUSTOMER OPENED · 12s AGO IRRIGATION DISTRIBUTOR · 41 SECONDS, START TO FINISH REAL WORKFLOW · DEPLOYED MARCH 2026 T+0s T+15s T+41s ✓
95%
Faster turnaround
3.2×
Close-rate lift
87
Quotes · first week
4 wk
To live
The setup

Half the week on quote admin.

The company sells irrigation parts to commercial farms across the Lowveld — sprinkler heads, drip line, valves, fittings. Their four sales reps were drowning. Each week brought around 85 enquiries, every one needing a custom quote: pulling SKUs, applying customer-specific pricing tiers, formatting a PDF, sending it back. Average time per quote: just under half an hour. Multiply that out and the team was spending roughly 40 hours a week — a full sales rep's worth of capacity — just typing quotes.

The bigger problem wasn't time. It was speed. Their fastest competitor was responding within an hour. Their median turnaround was around 4 hours. They were losing deals not on price, but on response time — and they knew it.

What we built.

We took four weeks to ramp. Week one: connected their Outlook to Quote Agent in read-only mode. Watched 30 enquiries flow through. Week two: ingested 18 months of past quotes and tuned the model on their actual pricing language, line-item conventions, and customer tiers. Week three: built the PDF template to match their existing quotes pixel-for-pixel and wired up the Telegram approval bot. Week four: ran a parallel pilot — every enquiry went to both the manual workflow and Quote Agent, and reps approved or rejected the AI draft. The agent's drafts were accepted unedited 74% of the time by week's end.

The first week.

Week one of go-live: 87 quotes sent. Same week the year before: 12 quotes sent in the same time bucket. Median turnaround dropped from 4 hours to 12 minutes — and most of those 12 minutes were the rep's commute to where they could actually look at their phone.

"We thought we'd lose control by letting AI draft the quotes. Instead, we got it back. The reps spend their day actually selling now — phone calls, site visits, follow-ups — instead of formatting PDFs." — Sales manager, Mbombela irrigation distributor

Three months in, close rate on quoted enquiries had moved from around 18% to roughly 58%. Some of that is faster response, some is consistency (no more typo'd line items or wrong banking details), and some is the simple fact that nothing falls through the cracks anymore — every enquiry that hits the inbox gets a quote, automatically.

What's next.

We're now scoping a stock-availability integration with their warehouse management system, so the AI can flag low-stock items at quote-time. After that, automated follow-ups for unanswered quotes after 48 hours. Same model, same approval flow, just more eyes — and less manual work for the reps.

Your before / after

Want this on
your real catalogue?

30-minute demo. We'll connect a test inbox, ingest a slice of your past quotes, and run live enquiries through the Telegram approval flow. No commitment.