Customer success stories
Every story below shows the specific problem, what we built, and the numbers before and after, backed by real screenshots wherever we have them.
| LPO_NUMBER | SUPPLIER | STATUS | DATE |
|---|---|---|---|
| 267912 | Melvins Marsh International Ltd | ● OK | 2026-08-12 |
Melvins Marsh International
FMCG Distribution · Kenya
Kenyan tea and food manufacturer/distributor, the makers of Melvins Teas.
Before
Every LPO was processed by hand: open the email, download the attachment, read each line, retype it one by one, then post it. That was a full day of work per batch. Some customers send handwritten LPOs; one customer alone aggregates ten to twelve PDFs into a single email. Volume runs 30 to 100+ LPOs a day. The backlog broke sales, accounting, and reporting downstream.
What we did
An automated LPO processing pipeline reads incoming emails and attachments (including scanned/handwritten formats), extracts line items, and posts them directly, running overnight and through weekends with minimal human intervention.
7-11%
revenue impact
3,100+
hours saved so far
3,590
LPOs processed, Jan–Aug 2026
Mr. Rooter Plumbing of Ottawa
Home Services · Ottawa, Canada
Residential and commercial plumbing company serving Ottawa-Gatineau.
Before
After-hours calls went unanswered or straight to voicemail, costing the business emergency-service jobs and putting admin staff under constant phone pressure during business hours.
What we did
An AI voice agent answers every call in under 2 seconds, 24/7, and automated dispatch notifications keep technicians and jobs moving without manual phone coordination.
<2 sec
average call answer time
1,500+
calls handled in 3 months
$400K+
in revenue protected
Beauty Square KE
Beauty Retail · Nairobi, Kenya
Nairobi beauty retailer running over 1,000 products across two branches.
Before
Before June, the inventory system was the owner. The bestsellers he watched by hand rarely ran out, but nobody can watch a thousand products: good-but-quieter sellers sat empty 17-27% of every week and nobody knew, because nothing measured a stockout. In a typical week about 16.6% of all bestsellers were out of stock while customers still wanted them, worst weeks passed 35%, ordering ran on instinct with no check against stock already on the way, and dead stock froze cash in silence. No number existed for what any of it cost.
What we did
A daily inventory system replaced hand-watched reordering. Every shelf is counted daily; Monday at 6am one email arrives with the six-week trend, top sellers and steady sellers split out, what's empty with nothing on the way, a costed buy list, and what to move from the warehouse. Three decisions, done by 6:15. The buy list only counts days a product was actually on the shelf, ignores one-off bulk buys and spike weeks, and subtracts what's on hand and already coming so it never orders twice.
16.6% → 11%
bestseller stockouts, weekly average (Jun-Jul)
0 weeks
above the 20% danger line since tracking began, vs. 8 of 23 weeks before
~23%
of dead stock cleared in the first tracked month (184 → 150 slow-moving SKUs)
Photography & Videography · Nairobi, Kenya
Photography and videography studio serving Nairobi, Kisumu and Juba.
Before: Sales inquiries and bookings were handled manually, replying to each one by hand, which meant missed opportunities whenever a message came in outside working hours.
What we did: Deployed an AI chatbot that handles sales inquiries and bookings end-to-end, including overnight and on weekends, so the team only steps in for exceptions.
Waste Management · Southern California, USA
Dumpster rental company serving LA, San Bernardino and Orange County since 2001.
Before: Manual data entry and lead-attribution gaps despite running active campaigns across Google, Bing, Facebook and Yelp, with no clear view of which channel actually drove a booking.
What we did: Built a Pipedrive-centered CRM automation: unified lead capture from web forms, CallRail phone tracking, Facebook, Google My Business and Yelp; automatic lead-to-deal creation with source attribution; round-robin lead routing to reps; automated SMS follow-up sequences via MailerLite; and custom dashboards for ROI attribution.
Personal Injury Law · Oklahoma City, USA
12-person personal injury law firm, 15 years in operation.
Before: Document generation took 25+ minutes per document (4-6 documents per client), intake ran over 2 hours of admin work before any case action began, and manual records-request tracking led to missed follow-ups.
What we did: Built an AI automation system ("Vikki") that generates legal documents, including LOPs, HIPAAs, and engagement letters, in 5 minutes, automates records-request follow-up at week 3 and escalation at day 35, and consolidates admin work to one person handling exceptions.
25 → 5 min
per document generated
2 hrs → <30 min
intake time per case
+14 cases
added in the first 2 months
JC, Paralegal, PI Associates
Marketing / UGC · Ecommerce & DTC
Creative performance agency specializing in UGC and psychology-driven social content for ecommerce and DTC brands.
Before: Grew mainly through inbound, referrals and paid traffic. Cold outreach was generic, and manually researching each prospect across Instagram, TikTok and Meta Ads to personalize outreach wasn't viable without hiring an expensive SDR team.
What we did: Built an automated outbound engine orchestrating Google Sheets, RapidAPI, Apify and the Meta Ads Library: it researches each prospect, scrapes their social data and ad performance, identifies competitors, and generates personalized audit-style emails, and one operator loads a batch and triggers the workflow.
~95%
less manual work per prospect
1,500–2,400
personalized emails sent per day
1 → 3-5
headcount leverage vs. hiring SDRs
Alessandro Wyatt, LoudLion
More teams we've worked with
Book a 30-minute audit. We'll map your processes, show you where you're losing time and money, and tell you honestly whether automation will help, or if you just need a better spreadsheet.
