Proof, not adjectives

Six problems I was handed — and what changed.

If you're hiring an operations or AI leader, you don't need more adjectives — you need to know what happens after I start. Here are six real situations, the moves I made, and the numbers that moved. The product screenshots are live; click into any of them.

$170K recovered on freight 72h → 6min order processing 10+ systems running unattended 7 entities onto one ERP $0.5M → $3M channel built Live products shipped

01 · Operations × Cost recovery

$170K hiding in the freight bill.

Situation
A $150M+ portfolio ships ~30,000 DTC parcels a month on roughly $5M of annual carrier spend. Nobody was auditing the invoices against the contracts.
Action
I stood up a carrier-audit initiative — reconciling billed charges against negotiated rates, line by line — and ran it alongside a US + Canada 3PL rebuild: RFPs and rate negotiations across UNIS, GoBolt and a multi-vendor field.
Result
$170K in recoverable billing errors surfaced on ~$5M of spend, plus a rebuilt 3PL network.

What this means for youI read the P&L like a system with leaks — and I bring a method to find them.

$170K
recoverable billing errors found
~$5Mparcel spend audited
30k/moDTC shipments
72 hrs→6 min
>90%of orders automated
1stEDI hire + pipeline

02 · Operations × Automation

Seventy-two hours to six minutes.

Situation
Dropship and instock orders were processed by hand — a 72-hour cycle that capped how fast the business could sell across its channels.
Action
I hired the first EDI developer, built the integration pipelines end to end, and led the SellerCloud + Kibo integration behind them.
Result
Over 90% of dropship + instock orders automated; processing time cut from 72 hours to under 6 minutes.

What this means for youI turn the manual bottlenecks that cap growth into pipelines that don't need me.

03 · Operations × AI

Ten systems that run before anyone logs in.

Situation
Operations spans 40+ channels and five divisions. The failures that cost money — a freight exception, a missing PO, a non-compliant listing — surface faster than any team can watch for them.
Action
I built a live suite of 10+ production systems with Claude Code: a freight-issue scanner across 1,000+ international shipment rows, a missing-PO monitor on 900+ POs, weekly SKU-by-SKU catalog-compliance reporting, PO-cancellation early warnings, and carrier-credential monitors.
Result
All of it runs unattended before business hours — including through a full week I was on leave.

What this means for youI don't add headcount for vigilance — I build the system that watches, and escalates only what matters.

See the control room
10+live systems
1,000+shipment rows scanned
900+POs monitored
24/7unattended
7 → 1
legal entities onto one ledger
115backward-planned tasks
QB → Intacctmigration

04 · Operations × Finance

Seven entities, one reporting spine.

Situation
Seven legal entities on QuickBooks, no clean bridge between operations and finance, and a mandate to move to Sage Intacct.
Action
I'm the program lead for the migration — a 115-task, backward-planned build from go-live backward, covering chart of accounts, master data, banking, and an ops-finance reporting bridge.
Result
A single reporting spine across all seven entities, with operations and finance reading from the same source.

What this means for youI run the cross-functional programs most ops leaders would rather hand off.

05 · AI · Products shipped

I productized the operator.

Situation
Operator knowledge is only leverage if it leaves your head. I wanted the method to run without me — for my teams and as real products.
Action
I shipped live software: the Amazon Sales Strategist (runs a real operator's audit method on a seller's own data), ApplyVerified (a job-search OS + applier marketplace on Cloudflare), and the Income Accelerator Coach (a full AI career coach).
Result
Live products with real users — ApplyVerified alone: 10 sprints, 893 tests, 11 database migrations.

What this means for youGive me a repeatable problem and I'll hand you a product, not a deck.

06 · Amazon × Business intelligence

Shelf noise, turned into Monday decisions.

Situation
A brand can be "on Target.com" and still be invisible — losing the organic result, out of stock in stores, slow to ship. Dashboards show it; they rarely say what to do.
Action
I built a Target.com shelf-health report: search share-of-voice, price & ratings, store availability and delivery speed across 50 stores in 10 metros, 18 keywords and 60 products.
Result
Decisions, not dashboards — exactly which keywords are won, which pages are broken, and where the fulfillment gap really is.

What this means for youI ship the analytics that tell an operator what to do on Monday — not another chart to admire.

Open the live report

Hiring for ops or AI? Let's talk specifics.

Tell me the fire. I'll tell you how I'd put it out in the first 90 days — and show you where I've done it before.