Cold-calling carriers on a specific lane is one of the most repeatable, high-volume outbound tasks in brokerage. A rep building capacity for an upcoming customer commitment might cold-call 80 to 150 carriers across a few days. Most won't pick up. Of those who do, most won't have the right truck. Of those who do, most won't agree on rate. The rep ends with maybe 4 to 8 carriers added to active rotation for the lane.
This article walks through exactly how an AI does the same work. Not as a marketing pitch but as a technical breakdown of the steps. By the end you'll know what the AI is doing under the hood, where the leverage comes from, and why this is one of the highest-ROI use cases for a brokerage.
What "cold-calling for a lane" actually means
Set the context. A brokerage has a customer commitment for, say, 12 loads per week of dry van Dallas to Atlanta. The brokerage's existing carrier rotation covers 7 to 9 of those. The rep needs to expand the rotation by 4 to 6 carriers who can run the lane reliably.
The work involves:
- Generating a list of candidate carriers (DAT, Truckstop, internal database, MC lookups)
- Calling each candidate to gauge interest, equipment, lane coverage, and rate
- Filtering down to qualified carriers
- Onboarding the qualified ones into the rotation
Manually, this is two to three days of operator time spread across phone calls, voicemails, follow-up emails, and document chase. AI compresses it to a few hours.
Step 1: Building the candidate list
The AI starts by generating a candidate list. The inputs:
From your network
Carriers who've previously hauled this or adjacent lanes. Carriers who have equipment and capacity that match. Carriers ranked by historical reliability, on-time performance, and price posture.
From load board data
Carriers actively posting capacity in the lane region. Carriers searching for loads in matching equipment. Carriers with recent activity on Dallas-Atlanta or adjacent corridors.
From external sources
Carriers in your TMS that haven't been called recently. Carriers in MC databases who match the equipment and authority requirements.
The AI applies filters: equipment match, MC age (avoid MCs younger than six months unless explicitly permitted), insurance currency, safety scores above your threshold, fraud signal screening. The output is a ranked list of typically 100 to 300 candidates.
Step 2: Carrier scoring
Before calling, the AI scores each candidate. The scoring weighs:
Lane fit
How often does this carrier run Dallas-Atlanta or close adjacents (Dallas-Charlotte, Houston-Atlanta, Dallas-Memphis-Atlanta multi-leg). High lane fit = high prior probability of yes.
Equipment match
Dry van, 53 foot. Some carriers run mixed fleets. The AI flags whether the carrier has dedicated dry van capacity or whether they're occasionally running it.
Performance history
If the carrier has hauled for your brokerage before, the AI looks at on-time pickup, on-time delivery, claims history, and dispute frequency. New-to-network carriers get the benchmark from MC public data.
Price posture
Does this carrier typically negotiate up sharply, or accept first offer? This affects sequencing decisions later.
Capacity signal
When did this carrier last accept a load, and where? A carrier whose last booking was 200 miles from Dallas tomorrow is a high-probability candidate. A carrier whose last booking was in California three days ago is low-probability.
The output is a sorted list with priority scores. The top 12 to 24 carriers get the first wave of calls.
Step 3: The first wave
The AI dials in parallel. Six to twelve simultaneous calls. Each call uses a voice model trained on real broker-carrier conversations and tuned to your brokerage's tone.
The opening
Conversational, not robotic. Something like:
"Hey, this is Sam with Northstar Logistics. I'm trying to build out some capacity on Dallas to Atlanta for a customer of ours. Got a regular dry van load, 12 a week. Wondering if you're running that lane?"
The opening signals three things: the AI is from a real broker, it has real freight, and it's asking a specific question.
Common responses
The AI handles the common reply patterns:
- "What kind of pay?" The AI gives the lane average and asks if that works as a starting point.
- "When?" The AI gives the day-of-week pattern and the volume.
- "We don't run that lane regularly." The AI thanks them and asks if they'd consider it for the right volume.
- "Send me the details." The AI commits to email and follows up immediately.
- "Don't call again." The AI logs the do-not-contact and ends the call.
Average call length: 90 seconds for a no, 4 to 6 minutes for a productive conversation.
Step 4: Filtering the responses
Out of the first wave of 12 calls, expect roughly:
- 4 to 6 don't pick up. Voicemail left, follow-up email queued.
- 2 to 3 pick up but the lane isn't a fit. Logged as not interested.
- 2 to 3 pick up and have lane fit but want more details. Email follow-up sent with rate parameters and next step.
- 1 to 2 pick up, have lane fit, and are immediately interested in talking specifics.
The AI captures structured outcomes from each call. Equipment confirmed, lane interest confirmed or denied, rate posture, contact preference (voice vs email going forward), follow-up timing, and any specific notes the carrier mentioned (driver count, home base, lane preferences).
This data becomes part of the carrier's record for the next time the brokerage has freight on a related lane.
Step 5: The follow-up wave
Carriers who didn't pick up get follow-up calls 4 to 8 hours later. The AI rotates timing to catch carriers at different points in their day. After two failed attempts, the AI switches to email or text and the cadence stretches to once every 1 to 2 days for a week.
Carriers who picked up but wanted email follow-up get the email immediately, with specific rate parameters, lane details, and a call-to-action to confirm interest.
Persistence without nagging
The AI maintains call frequency rules. No more than 2 calls in a 24-hour window. No calls before 7 AM or after 9 PM in the carrier's local time. Stop attempting after 5 unsuccessful contacts. These rules are configurable by brokerage.
The point: the AI is persistent but not annoying. It gives carriers the same respect a working operator would.
Step 6: Negotiation and onboarding
For carriers who want to engage on rate, the AI shifts into negotiation mode. The conversation pattern:
Setting the rate range
The AI opens with the brokerage's target rate (the floor). If the carrier accepts: book.
If the carrier counters: the AI compares the counter to the walk-away limit set for the lane. If counter is within range, the AI counters back at the midpoint. If counter is above the walk-away, the AI thanks them, says we'll keep them in mind for higher-rate lanes, and ends the call cleanly.
Multi-turn negotiation
Most negotiations close in 2 to 4 turns. Carrier: $2.40. AI: $2.18. Carrier: $2.32. AI: $2.25. Carrier: deal. Total time: 4 minutes.
What the rep sees
The rep doesn't manually do any of this. They see a dashboard:
- Total candidates called: 247
- Total connected: 89
- Total qualified: 22
- Total agreed to onboard: 8
- Average call duration: 1 minute 47 seconds
- Total operator time spent: 0
- Total AI time elapsed: 18 hours
Compare to the manual version: 2 to 3 days of operator time, 80 to 150 carriers called, 4 to 8 added to rotation. The AI processes 2x to 3x the volume in 1/40th the operator time.
The technical pieces under the hood
A few details that determine whether the AI does this well or poorly.
Voice latency
Round-trip latency from carrier speaks to AI speaks back has to be under 800 milliseconds. Above that, the conversation feels off. The voice agent uses streaming inference to keep latency low even on multi-turn conversations.
Interruption handling
When the carrier interrupts mid-sentence, the AI stops talking, listens, and responds to what they said. This is harder than it sounds and is one of the things that separates production-grade voice AI from 2022-vintage IVR.
Tone matching
The carrier has a tone (friendly, terse, professional, casual). The AI mirrors it lightly. A carrier who opens curt gets a more direct AI. A carrier who opens warm gets a more conversational one.
Network awareness
The AI recognizes when carriers ask for things that need network context: "What's the deadhead from where I'd be?" The AI pulls the carrier's recent location from the data and answers specifically.
Call recording and transcription
Every call is recorded and transcribed. The brokerage gets full transparency into what the AI said and what the carrier said. This is essential for quality monitoring, compliance, and continuous improvement of the voice model.
What this enables
Three things change for the brokerage when this becomes routine.
New lane buildouts move from days to hours
When the customer wants a new lane, the brokerage doesn't need to spend a week of operator time building capacity. Lane expansion stops being a hiring decision.
Carrier database becomes alive
A typical brokerage has thousands of carriers in its database. Most haven't been called in over a year. The AI keeps the database alive by routinely calling and re-engaging carriers who've gone dormant.
Rep time goes to relationships
The operator who used to spend two days cold-calling for a lane now spends those two days on customer reviews, lane strategy, and relationships with the top carriers in the rotation. The work moves up the value chain.
Common questions
Will carriers know it's an AI?
Some will. Most won't, on a first call. Of those who recognize it, most don't care because the conversation is productive. A small percentage prefer humans and will tell you so; the AI logs the preference and the next contact gets routed to a human rep.
What about Do Not Call lists?
The AI respects DNC at the carrier level (the brokerage's internal DNC list) and at the federal level. Cold-calling carriers in their commercial capacity is generally permitted but TCPA rules apply for some use cases. The AI ships with compliance defaults that fit common interpretations.
What if the AI says something wrong?
Calls are recorded and reviewed. Any error becomes training data. Errors are rare on the routine flow but more common on edge cases (unusual carrier questions, off-script topics). The AI hands off to a human when it encounters something it can't confidently handle.
How Ten8 runs this
Ten8 ships the AI cold-calling capability as part of the core platform. The carrier scoring, voice agents, onboarding workflow, and dashboard are all included. The rep team configures the lanes they want capacity for, sets the rate parameters, and the system runs the campaign in the background.
If you want to see this on your own freight, book a demo. We'll set up a lane-buildout pilot. Integration takes about two weeks.
Keep reading
What is an AI freight broker?
An AI freight broker is a platform that runs the operational layer of a brokerage end to end. Here's what that means in practice, and how to evaluate one.
Brokers guide to AI agents in freight
AI agents in freight aren't one thing. There's voice, email, chat, scoring, and orchestration. Here's a working taxonomy of the agent layer and how each piece fits a brokerage.
AI dispatch vs traditional TMS: what actually changes
Traditional TMS platforms digitize paperwork. AI dispatch systems make decisions. Here's why that distinction matters more than most brokers think.
