Case study — shipping

Invoice-processing bot: breakeven inside three months

A shipping company was spending time and money processing supplier invoices, with rework eating into the team's capacity on exceptions and mismatches. We scoped, built and deployed three targeted microbots — built with our engineering partner, Yodatech — against the invoice-matching workflow.

63%
Cost improvement
>99%
Rework reduction
2.2 mo
Breakeven
6→2
FTEs on the process

Cost and benefit

$34,200 of programme cost against $190K of annual run-rate saving

ItemPriceBasisTotal
Development cost$9,000Per process, fixed × 3$27,000
Application support$200Per process / month × 36$7,200
Ad-hoc requests$180Per person-day, as neededAs incurred
Total$34,200
Before
$300K / yr
Annual saving
−$190K
After
$110K / yr

6 FTEs on the process cut to 2, plus $10K in maintenance. Pricing assumes 12 months of application support after go-live, renewable; excludes application licence and hardware, which the client holds directly.

Why it matters

Microbots remain the foundation — rule-based, repetitive work: data entry, extraction, file manipulation. The AI agents we build now sit on top of this layer, not instead of it, and this is the discipline they inherit: scoped, priced, and measured against a baseline before and after.

Start with your own assessment →