How I think about building products. These aren't theories—they're lessons earned through real work at Loadsmart, applied to how I approach product strategy now.
The insight: The real bottleneck is rarely where you think it is. Half measures don't work.
When we built Parsers at Loadsmart, the obvious problem was "quote generation is slow." So we built automation to generate quotes faster.
But the data told us something different. Semi-automated quotes (parse request → generate quote) converted at 3-7%. Full-automated quotes (parse → generate → submit back to customer system) converted at 20-27%.
The actual bottleneck wasn't our speed. It was the customer's friction—they had to manually copy our quote into their system. That single step killed the win rate.
The lesson: Don't optimize the visible process. Understand the full loop. Close it.
How I apply this now: When I think about building agentic systems in logistics, I obsess over the full workflow, not just the AI's decision. The agent can be brilliant, but if the output doesn't integrate seamlessly into the human's workflow, adoption dies.
The insight: Constraints force creativity. Ship fast. Rebuild later if you need to.
When we launched LTL at Loadsmart, the platform was built entirely for Full Truckload—point-to-point routing, no freight classes, no accessorials, no terminal-based logic.
The tempting move: Rebuild the platform from scratch with LTL-first design.
What I actually did: Adapted the existing FTL system. Added LTL as a mode. Worked within the constraints.
Why? Because perfect architecture takes time. And in a scaling business, time to market matters more than perfection. We went from 50 loads/month to 3,000 loads/month by working within the system, not waiting for a rebuild.
The lesson: Constraints aren't obstacles. They're where good product thinking happens.
How I apply this now: I'm comfortable shipping imperfect solutions quickly and iterating. I'd rather have a scrappy agentic system that works with real data than wait for the "perfect" architecture. Real customers always reveal what actually matters.