We're Automating Drug Test Scheduling with AI — Here's What That Actually Looks Like

The Problem Nobody Talks About

Every day, compliance companies like ours coordinate drug and alcohol tests for commercial drivers across the country. A single pre-employment test involves collecting driver information, verifying what type of test is needed, finding a nearby collection site, placing the order with the right testing vendor, sending the driver their authorization to walk into a clinic, and then tracking the result.

Multiply that by hundreds of drivers a week, across multiple test reasons — random, post-accident, reasonable suspicion, return-to-duty, follow-up — and you’ve got a full-time job that’s mostly just moving data between systems.

That’s exactly the kind of work AI should be doing.

What We’re Building

We’re building a voice-powered AI agent that handles the entire scheduling process end-to-end. A motor carrier calls in, the agent picks up, collects the driver’s information conversationally, determines the right test type, places the order through our testing vendor’s API, and the driver gets their clinic authorization — all without a human touching it.

The key insight was that this isn’t a chatbot problem. It’s an integration problem. The hard part isn’t making AI talk — it’s connecting the voice agent to production APIs, handling the dozen different test types with different requirements, routing to the right vendor based on cost and coverage, and making sure every order meets DOT compliance standards.

Why Voice, Not a Portal

We could have built a web portal. Most companies do. But our customers are truck drivers and small fleet owners. They’re on the road. They’re used to picking up the phone. A portal adds friction. A phone call is natural.

The voice agent doesn’t just answer questions — it takes action. It creates the applicant record, submits the order, and triggers the scheduling workflow. By the time the call ends, the driver has what they need to walk into a clinic.

The Infrastructure Nobody Sees

The unglamorous reality of building AI systems is that 80% of the work is infrastructure. Getting production API access. Setting up secure connections with fixed IP addresses. Building proxy layers so cloud-based agents can talk to IP-restricted vendor systems. Configuring authentication. Pulling product catalogs. Testing end-to-end in production environments.

None of that is “AI” in the way most people think about it. But without it, the AI has nothing to do. A voice agent that can’t place a real order is just a fancy answering machine.

What I’ve Learned So Far

Three things stand out from this build:

Start with the integration, not the AI. We proved our API connections worked before we wrote a single line of voice agent logic. If the backend doesn’t work, the frontend doesn’t matter.

Security can’t be an afterthought. When you’re handling personal information for DOT-regulated drug tests, you don’t get to cut corners. Dedicated IPs, encrypted connections, secret management — all of it matters from day one.

AI implementation is iterative. We didn’t try to build the whole thing at once. We started with a Documents Agent that handles simpler tasks — retrieving invoices and resending emails. That taught us the architecture patterns we’re now applying to the more complex scheduling workflow.

What’s Next

We’re weeks away from our first live test order through the AI agent. After that, we’ll add routing logic to choose the most cost-effective testing vendor for each order, set up result notifications, and connect the whole thing to our CRM and billing systems.

The goal isn’t to replace our team — it’s to free them from repetitive coordination work so they can focus on the compliance expertise that actually requires a human brain.

I’ll share more as we go live. If you’re running a compliance operation and thinking about AI, feel free to reach out. The path is more achievable than you think — it just requires building the right infrastructure first.

Sarah Hope is the CEO of Vertical Identity, a DOT drug testing consortium serving 5,000+ motor carriers nationwide. She writes about implementing AI in regulated industries at sarahhope.co.