The breaking point.
The TMS that should have existed.
How one frustrated dispatcher decided to build the AI-powered TMS carriers actually needed — not another piece of broker software.
Mark Tishkun, Founder & CEO, never set out to build a software company. He was running dispatch and operations for a trucking carrier, dealing with the same chaos every operations manager faces every day: mountains of paperwork, nonstop problem-solving, and software that seemed determined to make the job harder instead of easier.
The TMS they relied on was a constant frustration. Every rate confirmation required 5–10 minutes of manual data entry. Assigning a driver often came down to who picked up the phone first, not who made the most operational sense. And every load felt like starting from scratch — no intelligence, no memory, no learning from past decisions.
Mark quickly realized the problem:
The people building transportation software had never actually run a trucking operation. They were building tools for brokers and 3PLs — full of dashboards, reports, and complicated workflows — but not for the carriers and operators who just needed something streamlined, fast, and intuitive.
Mark didn't want "fancy."
He wanted software that simply didn't waste time.
And when he couldn't find it, he started building it.
"I watched my best dispatcher spend 60% of her day typing information from PDFs into form fields. Meanwhile, AI companies were processing millions of documents per second. Something was deeply wrong."
The math didn't add up
Let's be real about the numbers. A good dispatcher can handle maybe 20–30 active loads at a time. Why? Because they're buried in administrative work that has nothing to do with actual dispatching.
Here's what a typical day looked like:
- → 30 minutes per load just on data entry
- → 15–20 calls to find available drivers
- → Zero learning from past assignment decisions
- → Constant interruptions for status updates
- → Manual invoice generation at month-end
Meanwhile, the software they paid thousands of dollars per month for didn't do anything intelligent. It was just a fancy database with forms. No automation. No intelligence. No learning.
The AI moment
In 2023, Mark watched GPT-4 extract structured data from a messy document in seconds — perfectly, effortlessly, with zero human input. That was the moment it clicked for him:
This is the technology that should be powering a modern TMS… not a system built on 1990s architecture.
He started experimenting. First, he built a simple OCR tool to read rate confirmations. Then he created an early model that could predict which drivers were the best fit for specific lanes. When Mark shared these prototypes with other dispatchers, the reaction was identical every time:
"Why doesn't our TMS already do this?"
That's when he realized the truth:
The entire TMS industry was built on outdated assumptions about what software could do.
AI had completely changed the game — but almost no one in trucking had noticed yet.
Building Atlas Command
Mark decided to build the TMS he always wished existed, not something slightly better but something fundamentally different. It would be AI first at the core rather than AI added as an afterthought, and designed for the people who actually move freight every day: carriers, dispatchers, and operators, not the brokers who shuffle paper.
The core principles were clear from day one:
1. Eliminate stupid work. If a computer can do it, the computer should do it. Humans should focus on relationships and strategy, not data entry.
2. Learn from every decision. Every load assignment teaches the system something. Your operational intelligence should compound over time, not reset with every new dispatcher.
3. Built for speed. Dispatchers don't have time for slow software. Everything should be instant, or it's broken.
4. Actually solve the problem. Don't add features because they look good in demos. Add features because they eliminate real operational pain.
The journey so far
Managed dispatchers for a regional carrier. Watched talented people waste their days on manual data entry instead of actually managing operations. Knew there had to be a better way.
Built an OCR tool that could read rate confirmations in seconds. Showed it to dispatchers. They couldn't believe legacy TMS platforms didn't have this. Neither could Mark.
Realized the entire TMS market was ready for disruption. AI had fundamentally changed what was possible, but nobody in trucking had built an AI-native system yet. Time to build it.
2,500+ loads processed in production, ~97% AI accuracy, and hours saved per dispatcher per day. The product worked. Now it was time to scale.
There are hundreds of thousands of trucking companies still using software from the 1990s. Atlas Command is here to give them something better — something that actually works for carriers.
Why now
AI has reached a tipping point. What was impossible a few years ago is now commodity technology. Document processing, pattern recognition, predictive modeling — all of it is mature enough to build a production system on.
But here's the thing: the technology is only half of it. The other half is understanding the problem deeply enough to build the right solution. You can't build great trucking software without having run trucking operations.
That's Atlas Command's advantage. The team isn't made up of outsiders trying to "disrupt" an industry they don't understand. They're insiders who got fed up with bad software and decided to build something better.
What's next
Atlas Command is in private beta with select carriers. The AI is learning. The product is working. Dispatchers are saving hours every day. Now the team is ready to scale.
This is just the beginning. Every trucking company deserves software that actually helps them, not software that gets in their way. Atlas Command is going to make sure they get it.
Be part of the story
Join the carriers who are already using AI to transform their operations — from manual data entry and guesswork to intelligent, self-improving dispatch.
Request beta access