
Case study · Project #1
A drilling
contractor
in Zimbabwe.
Our first pilot: a multi-site drilling and agriculture group. Their teams already ran the business on WhatsApp and email. We put agents inside that loop. Client anonymised; every example below is rebuilt with synthetic data.
The plan
90 days,
three phases.
DAYS 1–30
Connect & capture
WhatsApp groups and the stores inbox connected; requests and breakdowns captured automatically.
DAYS 31–60
See & chase
Ops board, daily digests and automatic chasing with supervisor approval.
DAYS 61–90
Reconcile & hand over
Stores, metres and fleet reconciled; a managing director's view on top.
Pattern · stores
Email in.
Order out.
A drilling contractor in Zimbabwe: site clerks email and WhatsApp stores requests. The STORES agent reads them, matches the store's list of items, merges duplicates and builds an order for approval.
01 · INBOX
Request arrives
“Need 4 × hydraulic hose 1/2in and 2 × 20L hydraulic oil for Rig B.” Email or WhatsApp.
02 · MATCH
Items matched
Agent maps free text to stock codes (example: HOS-12, OIL-20) and checks on-hand.
03 · MERGE
Duplicates flagged
Same request re-posted in the group? It is merged, not double-ordered.
04 · APPROVE
Order drafted
Manager taps approve; the order and a reply to the clerk go out.
Example workflow · synthetic data
Pattern · fleet GPS
Every truck,
on one map.
The FLEET agent reads GPS pings next to WhatsApp check-ins. Idle too long, off-route or late? It asks the driver, logs the reason and updates the ETA.
01 · PING
GPS + check-in
Example: Truck 4 idle 2h 15m near a border post.
02 · ASK
Driver asked
“Queue or problem?” in WhatsApp; voice reply is fine.
03 · LOG
Reason logged
“Border queue” attached to the trip; ETA moved.
04 · DIGEST
Fleet digest
Evening summary per vehicle: km, idle, incidents (example data).
Example workflow · synthetic data
Pattern · metres
Metres per rig,
per shift.
Drillers send their shift totals the way they already do. The METRES agent turns them into a daily output board per rig, with bits and downtime alongside.
01 · SHIFT
Shift report
“Rig A night shift 64 m, bit changed at 40 m, 1h standby for water.” (example)
02 · PARSE
Structured
Metres, bit, downtime reason and hours extracted.
03 · CHECK
Sanity checks
Missing shift or an outlier? Agent asks the driller before it hits the board.
04 · BOARD
Daily output
Metres per rig and shift, trend vs plan (example figures).
Example workflow · synthetic data
Pattern · breakdowns
Logged once.
Chased until fixed.
A breakdown message becomes an incident with an owner and a clock. The agent chases on schedule, escalates when it slips, and closes it when the fitter reports back.
01 · REPORT
Breakdown message
“Compressor on Rig C won't start.” Text, voice or photo.
02 · INCIDENT
Owner + SLA
Incident opened, workshop assigned, parts checked.
03 · CHASE
Follow-ups
Drafted reminders at the SLA mark; supervisor approves.
04 · CLOSE
Closed with cause
Fix, downtime hours and cause recorded (example).
Example workflow · synthetic data
What changed
Early signals.
Requests stopped getting lost in busy groups. Duplicate orders were caught before they were placed. Breakdowns had an owner and a clock. Management got one daily picture instead of scrolling chats.
Qualitative only · figures published once cleared with the client
Next step
See it on
your site.
Tell us how your teams report today. We will show you the loop running on it.