Recent work, by industry.

We name industries, not clients. We publish numbers only after the measurement window closes and the client signs off, so for now you'll see what we built and what we're measuring.

Small Business OS

Corporate catering

Orders and lead intake

What was broken

  • Every order came in by phone
  • No booking pipeline, so deals slipped when the line was busy
  • Language barriers slowed phone orders

What we built

  • Phone intake connected to the POS and CRM
  • A 24/7 voice agent that answers and triages catering leads
  • Live reporting on order values and lead sources

What we're measuring

  • Concurrent order capacity
  • Intake handling time
  • Lead-source visibility

Independent retail

Storefront and customer service

What was broken

  • The whole business ran through a marketplace
  • Vendor supply tracked by hand
  • "Where's my order?" questions all day

What we built

  • An independent storefront with vendor and inventory sync
  • An assistant for order-status questions that hands anything unusual to a person
  • Customer data owned by the shop, with dashboards

What we're measuring

  • Routine inquiry volume
  • Response time
  • Escalation rate

Custom design studio

Missed calls and follow-up

What was broken

  • Missed calls went cold
  • Weak local search presence
  • Follow-up depended on memory

What we built

  • A text-back agent that turns missed calls into CRM bookings, 24/7
  • Automatic feedback and review requests
  • Dashboards for margin and client value

Where it stands

  • Missed calls now get a text back and a booking path, 24/7
  • Results to follow after measurement and client sign-off

From the warehouse lab

These ran on test environments, not client systems. Nick shares them as he builds them on LinkedIn.

Trace Analyzer

Lab test

The problem

  • Only one person on the team could read a Blue Yonder trace

What it does

  • Takes a trace by email, searches the whole file and replies with the root cause and evidence

What we saw

  • A 23 MB, 173,000-line trace: the real cause sat 157,000 lines above a misleading final error. Found in under five minutes for about a dollar.

MOCA Dev Agent

Lab demo

The problem

  • Small WMS changes quoted at weeks of hours

What it does

  • Builds changes on typed MOCA commands, with tests and docs, for a person to review

What we saw

  • A working RF form in 10 minutes. The same form had been quoted at 80 hours.

Cut-Off Rescue Agent

Design

The problem

  • Orders near cut-off discovered too late to save

What it does

  • Finds at-risk orders, queries the cause, reads the ops inbox and briefs the supervisor

Status

  • Available as a scoped pilot after the assessment
Needs your approval

Want to be the next example?

Start with a free check. If we build something together, you decide whether your results ever get published.

Get a free leak check Send a trace for review