This is the story of how we told a client not to use AI — and saved them $340,000 in the process. It's also the story of why we named our company "Pragmatic."
A regional logistics company came to us with a clear request: they wanted an AI-powered route optimization system. Their fleet of 80 delivery trucks was growing, routes were getting more complex, and they'd heard that AI could reduce fuel costs by 15–20%.
The Assessment
We spent two weeks embedded with their operations team. We rode along on deliveries, sat in on dispatch meetings, and analyzed their routing data. What we found surprised everyone — including us.
The problem wasn't that their routes were suboptimal. The problem was that they were running the right routes at the wrong times. Their dispatch schedule hadn't been updated in three years, and it was based on traffic patterns that no longer existed. A highway expansion had completely changed the optimal delivery windows.
The Non-AI Solution
Instead of building an AI route optimizer (which would have cost $200K+ and taken 6 months), we recommended three changes:
- Updated the dispatch schedule based on current traffic data from Google Maps API — a simple script, not AI
- Reorganized delivery zones to match the new highway access patterns — literally a spreadsheet exercise
- Shifted two delivery runs to off-peak hours — a scheduling change that required zero technology
Total implementation cost: about $15,000 (mostly our consulting time). Total annual savings: $340,000 in fuel, overtime, and vehicle wear.
Why We Said No to AI
An AI route optimizer would have worked. It would have found the same inefficiencies and probably squeezed out an additional 3–5% in savings. But the cost-benefit didn't make sense:
- $200K+ development cost vs. $15K for the simple solution
- 6 months to deploy vs. 3 weeks
- Ongoing maintenance and data pipeline costs vs. a quarterly schedule review
- The simple solution captured 90% of the available savings
The remaining 10% of savings that AI could have added? It would have taken 3+ years just to break even on the development cost.
The Lesson
AI is a tool, not a strategy. The strategy is to reduce logistics costs. AI is one possible tool to achieve that — but it's not always the right one. Sometimes a spreadsheet, a schedule change, or a simple script is the pragmatic answer.
We could have sold them the AI project. It would have been a $200K engagement for us instead of a $15K one. But that's not how we operate. Our reputation is built on giving honest advice, even when it costs us revenue.
When AI IS the Right Answer
To be clear, we're not anti-AI. We build AI solutions every day. But AI is the right answer when:
- The simple solutions have already been exhausted
- The problem involves pattern recognition at scale that humans can't match
- The data volume and complexity justify the investment
- The ROI math works even with conservative estimates
For this logistics company, AI might be the right answer in 2–3 years when their fleet doubles and the routing complexity genuinely exceeds what simple scheduling can handle. When that day comes, we'll be ready to build it.
Not sure if your problem needs AI or something simpler? Book a free assessment — we'll give you the honest answer.