Work that moved a number
Four engagements across logistics, ecommerce, SaaS and executive search. Different problems, same pattern: audit first, build fast, measure honestly.
Nordvale Logistics
An AI voice agent that took the night shift off dispatch
Nordvale runs regional freight across Ontario and Quebec. Their dispatch team was losing entire days to inbound calls asking the same four questions about load status, pickup windows, proof of delivery and driver ETAs.
Results
The problem
Dispatchers were fielding roughly 210 inbound calls a week. Most of them were status checks that the data already answered. Anything after 6pm went to voicemail and got handled the next morning, which meant carriers and customers were waiting overnight for information the system already had.
What we did
- Mapped every inbound call type over two weeks and scored each one for automation potential.
- Built a voice agent connected to their TMS so it could read live load status, ETAs and delivery confirmations.
- Trained it on their tone, their terminology and their escalation rules, including when to pull a human in.
- Added an after hours mode that books callbacks straight into the dispatch calendar.
- Wrote every call back to the CRM with a transcript and a summary line.
We went from a first call on Monday to a working prototype that Friday. I have worked with four agencies before and none of them moved like that.
Solene Beauty
Cutting first response time from nine hours to forty seconds
Solene sells skincare direct to consumer across France and Benelux. Growth was healthy but support was drowning, and the checkout was quietly losing shoppers at the shipping step.
Results
The problem
A two person support inbox was averaging nine hours to first reply. Pre purchase questions about ingredients, shipping and returns were going unanswered long enough that the sale was already gone. Meanwhile checkout analytics showed a heavy drop between the cart and the shipping screen.
What we did
- Built a support agent trained on the full product catalogue, ingredient sheets, shipping matrix and returns policy in French and English.
- Connected it to the order system so it could answer "where is my order" without a handoff.
- Set clean escalation rules for anything involving a refund, a complaint or a skin reaction.
- Ran a UX audit on the checkout and rebuilt the shipping step around one decision per screen.
- Tested three versions of the cart page over six weeks and kept the winner.
The audit alone was worth the engagement. They found three places we were burning budget that our previous agency had missed for a year.
Kaido Fitness
Quadrupling organic traffic and getting cited inside ChatGPT
Kaido sells gym management software to independent studios. They were spending heavily on paid search with a CPA that made the unit economics uncomfortable, and their organic presence was close to invisible.
Results
The problem
Almost all pipeline came from paid, at a cost per acquisition that only worked on annual plans. Organic sessions were flat. And when their buyers asked AI assistants for gym software recommendations, Kaido never came up, while three competitors did.
What we did
- Rebuilt the site architecture around the questions studio owners actually ask, not around internal product naming.
- Shipped a comparison and alternatives cluster targeting the exact queries buyers use late in the decision.
- Implemented full schema markup, entity consistency and citable statistics pages built for generative engines.
- Earned placements on the review sites and roundups that AI assistants pull from.
- Rebuilt the two highest traffic landing pages and cut paid spend on the keywords organic had taken over.
Papaya rebuilt our site and our organic traffic quadrupled in seven months. What surprised me more was showing up inside ChatGPT answers for the terms we care about.
Marlowe & Finch
Shortlisting in eleven hours instead of six days
Marlowe & Finch place senior finance talent. Every search started with two consultants reading several hundred CVs, which was the least valuable and most expensive part of their week.
Results
The problem
A typical mandate drew 300 to 600 applications. Two consultants spent close to two full days per role on first pass screening. By the time a shortlist reached the client, the strongest candidates had often already accepted elsewhere.
What we did
- Built a screening agent that reads CVs against the specific criteria written into each mandate brief.
- Scored candidates on evidence rather than keywords, with a written justification for every score.
- Flagged borderline profiles for human review instead of silently rejecting them.
- Automated the first outreach sequence and interview scheduling for anyone above the threshold.
- Built a bias check into the review step and kept every decision auditable.
Screening used to eat two full days a week. The agent they built shortlists overnight and the quality went up, not down.
Your situation is different. That is the point.
Every one of these started with the same thirty minute call. Book yours and we will tell you what we would look at first.
Based in North York, Ontario. Working with clients across North America and Europe.