How we help logistics companies save $1M+ using AI (playbook + what actually sells)

Hey IH 👋

Sharing another set of learnings from the trenches. Last post I wrote about IoT + data engineering for GCC freight. This one is specifically about AI freight cost optimization — a category where the ROI is so clear, the sales cycle almost sells itself... IF you position it right.

The market setup

Freight is a huge, boring, high-margin problem:

Companies losing 15-25% of freight spend to inefficiency
Manual freight audits take 2-3 weeks
Detention/demurrage fees are massive ($250K+/year for mid-sized shippers)
Fuel volatility eating margins

The kicker: most freight teams STILL run on spreadsheets in 2026.

The offer that closes

We stopped selling "AI." Nobody buys AI.

Instead we sell:

"We'll cut your freight spend by 15% in 6 months or we don't get paid."

That's it. That's the pitch.

The proof point we lead with: one client (mid-sized retailer) went from $250K/year losses to $1.2M in savings in 6 months.

35% reduction in detention expenses
12% fuel savings
98% on-time delivery

That case study alone closes more deals than any deck.

The 5-pillar framework we sell against

Every AI freight engagement maps to one or more of these:

Route & mode optimization (10-25% savings per shipment)
Automated freight audit (weeks → hours)
Load consolidation (30% lower per-unit cost)
Dynamic pricing intelligence (15-20% reduction)
Carrier scorecards (better contracts)

We let the buyer pick the ONE they want to attack first. Small commitment. Fast win. Then we expand.

What we learned about pricing

Three pricing models we tested:

Fixed fee project — good margins but no compounding value
Time & materials — flexible but hard for buyers to commit
Revenue share on documented savings — hardest to sell but best for retention

The winner for enterprise deals: Fixed fee for pilot ($X), then revenue share on measurable savings beyond baseline.

Buyer risk = capped. Our upside = uncapped. Everyone's aligned.

The data problem nobody talks about

Every single engagement starts with the same problem: their data sucks.

Multiple ERPs that don't sync
15-year-old TMS with no API
Excel exports as "reports"
40% incomplete records

We now BUILD IN a mandatory data engineering phase before any ML modeling. Charge separately for it. It's 30-40% of engagement value and it's the highest-value work we do.

Lesson: if you're selling AI to traditional industries, price the data cleanup separately. It's not "prep" — it's the actual product.

Trends we're riding into 2026

Generative AI + NLP for supply chain — buyers can ask their supply chain questions in plain English
Green logistics AI — carbon compliance (EU CBAM) is a forcing function
Autonomous freight negotiation — early stage but coming fast

Full breakdown

I wrote up the technical pillars and case study details here:

Optimizing Freight Costs with AI in Logistics & Supply Chains

Questions for the community

Genuinely want to hear from other founders:

Anyone else selling "outcome-based" or revenue-share models? What percentage of your book is that vs fixed fee?
How do you structure the data cleanup work — bundled or separately priced?
For enterprise sales in traditional industries: what's your average sales cycle right now? Ours is 3-5 months and I'm curious if others are seeing longer/shorter.

Drop your experience below

Full disclosure: I'm with INTECH Group. We focus on AI/ML, data engineering, and digital transformation for logistics, ports, manufacturing, and supply chain. Always happy to trade notes with folks building in the same space.