The calculus around building your own tools and integrations has completely changed in the past few years. In this article, I'll review how we got here, what's changed, when it makes sense to build, and where I'd start if I wanted to develop the muscle of technical self-reliance.
Five Years Ago
Five years ago, it took an entire team working months or years to build a core piece of your toolbox. Once it was complete, the maintenance and complexity kept that team on long-term. If you wanted one-off integrations or automations, that might require yet another team member or a set of consultants. If you were a $100M+ freight broker whose core competitive advantage was systems and automation, that made sense.
Small brokers were generally stuck with off-the-shelf solutions and a lot of manual work, which forced them to scale headcount as they grew. If they wanted automation or integrations, they needed outside technical help to build it and keep it running. That worked for plenty of brokers, but the added cost eroded margins, so it didn't pencil out for most. That's how we got where we are today: tons of brokers are looking to get ahead with systems, technology, and automation. The only question is how to do it right.
Today
Today, pretty much any freight broker with one motivated individual can build real automation and integration for their own organization. AI is a genuine revolution, not just for your vendors. The fundamental low-cost tools now let you do much more yourself the fundamental low-cost tools let you do much more yourself. Some brokers are already using this to grow far more efficiently than their competitors.
How I'd Get Started
Let's imagine I was invited to lead technology at a new freight brokerage called Second City Logistics (an homage to Chicago's nickname). This brokerage has gotten off the ground and now wants to build automation and tools from the early days. Here's where I'd start:
I'd get an organization-wide subscription to a foundational model company's tool: Claude Enterprise, OpenAI Business, or Gemini Business. Every single person in the organization should have access. It matters far less which tool you pick (they're all good) than whether people are actually using it.
I'd create a Slack, Teams, or Google Chat channel to share ideas and feedback on how to use these tools to automate work.
Person by person, I'd connect that subscription to the key systems each person uses: email, chat, load board, TMS, and so on. Some connections may not exist yet (TMS is a common gap), so I'd focus on the ones that do. I'd use MCP or out-of-the-box connectors, but if you want tighter control, something like Zapier lets you connect systems securely and in a centralized way.
I'd teach team members how to collaborate with each other and with AI on common problems. In Claude, for example, you can build projects where multiple users work alongside AI in the same place. I'd build one on shipper pricing strategy, another on carrier sourcing standards, another on QBR prep. Over time, you'll use AI to build standardized templates you can apply across the organization.
Once these tools are fully available and collaboration has started, the seed of a more autonomous, AI-enabled organization has been planted. Next, I'd look at centralized integrations and automations.
Integrations & Automation
Getting high-quality integrations and automations running is harder than rolling out an enterprise AI tool. You'll need someone with some technical experience (some coding, but not a ton), a bit of infrastructure (somewhere for the integration to live), and the ability to iterate and maintain it, meaning it lives in someone's job description rather than as a side project. You'll also need a TMS with APIs and, ideally, event-based integration methods like webhooks. Once you have those pieces together, you're off to the races.
The first thing I'd do is build an integration between the TMS and a standard provider like Zapier. The TMS should be the center of your universe; integrations should generally run from the TMS out to other systems. While that was underway, I'd work with the team to identify the highest manual-lift items. Common answers are load building, POD collection, detention requests, and check calls.
If I already had a vendor covering one of these, I'd use it first. If not, I'd wire up Zapier to handle it. For load building, that means connecting email, processing the messages with an LLM, then formatting the output correctly to send to the TMS. Just like that, your first integration and automation is live. I'd test it, retry until it works at a high level, then roll it out across the team.
More brokerages should and will start to own their own integrations and automations. It's a great capability to have in house, it's more accessible than ever, and it will be a competitive advantage in the near future.
Core System Builds
The hardest build, and potentially the most valuable, is your own core systems, like a TMS. Unless you have a dedicated team, I'd advise against it, and the reason isn't obvious. It's easy to build a beautiful mockup that has exactly what you want. Getting everything to actually work well requires deep expertise and sustained focus. There's a reason the biggest AI companies in the world don't vibe code their own CRM (they use Salesforce) or HR system (they use Workday), and it isn't that they can't. It's that building those tools is a distraction from their core competency. The same is true for freight brokerage.
There's one other thing about vendors of core brokerage systems that's easy to overlook: they see across many freight brokerages. That means they're already solving problems you haven't encountered yet but likely will. Choose great vendors for your core systems. Look for ones that support integrations easily, use AI, and innovate at a fast pace.
There has never been a better time to become more technically self-sufficient as a freight brokerage, but it takes real effort to realize the potential and turn it into a lasting advantage.
Own more of the connective tissue. Buy the systems of record.