Agents for the Exceptions That Eat Your Day
Every mile documented. Every exception resolved. Every system answering as one.
We build agents for carriers, brokers and 3PLs. The TMS was never the problem. The check calls, the missing POD and the load that went quiet at 2am are the problem.
“Where is load 88214 and will it make the appointment?”
Status and stops
TMS
Current position
Tracking feed
Driver’s last update
Text thread, 40 minutes ago
Delivery appointment
Scheduling record
Running 90 minutes late. Reschedule request drafted for the scheduler.
Four systems asked at once. Nobody opened four tabs, and nobody made a phone call to find out.
Three of These Are Already Running
Not a pilot, not a proof of concept. Three named companies, published numbers, all of which you can check before you call us.
- Landstar: 90% less time spent searching. One portal answers questions that used to mean opening four systems, so the person on the phone stops hunting and starts answering. That is the Shipment Visibility Agent below.
- Bennett: 95% of invoice extraction automated, and a 70% faster processing cycle. Documents read, extracted and validated without a person retyping them, including the ones that arrive as a photograph of a crumpled page. That is the Document Processing Agent below.
- Carter, a Caterpillar dealer: routing efficiency up 30%. Dispatchers, drivers and technicians were on separate systems for routing, tracking and communication. One dashboard and a driver app later, updates reach the customer without anyone making a call. That is the Dispatch Exception Agent below.
Both are carriers, and we would rather say so than let you assume otherwise. For a broker or a 3PL the paperwork and the settlement are the same. What differs is who you chase, and that is scoping, not a rebuild.
Both are published case studies. We will also tell you on a call what did not work the first time, which is usually the more useful half of the story.
Getting It Built Is Easy. Getting It Through Peak Is the Job.
Freight does not stop for a deploy. Volume triples, a carrier’s API goes down, a receiver changes a form. The agent has to keep working or get out of the way cleanly.
It degrades, it does not disappear
When a feed goes down the agent says so and hands the load to a person with what it already knows. Silent failure is worse than no agent, because dispatch stops checking.
A known cost per load
We agree the cost per load or per document before the build, and meter and cap it per workflow, so the unit economics hold in the month you were busiest instead of surprising you.
It speaks to what you already run
TMS, tracking feeds, EDI 204s and 214s, the carrier portal with no API, the receiver who still wants a fax. We have built against all of it, including the scheduled file drops.
A person still commits
Agents draft, schedule and recommend. Tendering a load, releasing a payment and committing to an appointment stay with your people unless you explicitly decide otherwise.
Proven on your worst documents
The accuracy bar gets set on your paperwork before anyone commits, not on our benchmark set. The handwritten BOL and the faxed POD are the test, because they are the actual job.
Someone is up when you are
Monitoring, drift checks and the morning a partner changes their file format are ours. Your operation runs at 3am, so the thing watching the agent cannot keep office hours.
These are the same controls we publish openly, applied to an operation that never closes. See the governance framework in full →
Sixteen years building production systems, well before anyone called it agentic.
A firm with a bench, not a two-person shop that disappears after launch.
No offshore delivery. The engineers who build it are the ones on your calls.
We build in your environment. Your data does not leave it.
Bring us one workflow. We prove it pays, build it properly, control it so security signs off, and keep it running after everyone else has moved on.
Nobody buys ten agents. You buy one, it survives a peak, and the second one is an easier conversation.
Agents for the Load, the Paperwork and the Payment
High-frequency, document-heavy, exception-heavy work where the rules are known and the volume is punishing. Filter by where in the cycle it sits.
Track is the load and its paperwork while it moves. Process is the calling, the chasing and the answering. Settle is getting paid correctly, proving it later, and pricing the next one.
Shipment Visibility Agent
One question, every system answers. Track, trace, and ETA from a single interface.
OperationsDocument Processing Agent
BOLs, PODs, and rate cons read, extracted, and validated in seconds. Even the handwritten ones.
Back OfficeDispatch Exception Agent
Delays and disruptions flagged, triaged, and routed before the customer calls.
DispatchYard & Terminal Agent
Trailers, docks, and moves orchestrated by signal, not clipboard.
Yard OperationsCarrier Communication Agent
Check calls, appointment scheduling, and load updates handled by voice and text.
Carrier RelationsCustomer Service Copilot
Shipper questions answered from live shipment data with full context, in any channel.
Customer ServiceCompliance & Audit Agent
DOT files, insurance certs, and audit documents extracted and always inspection-ready.
Safety & ComplianceInvoice & Settlement Agent
Carrier settlement matched, reconciled, and disputed with evidence attached.
FinanceCapacity & Demand Forecasting
Tomorrow's volume predicted from today's signals, lanes priced accordingly.
Pricing & CapacityKnowledge & SOP Search
Every SOP, tariff, and procedure answered in plain English with citations.
Enterprise / FrontlineNothing in that stage yet.
Three Sensible Places to Start
Not the biggest theoretical payoff. The ones where the workflow is bounded, the data already exists, and you can prove it inside a quarter.
Document Processing
The one with the hardest number behind it. Paperwork arrives in the same shapes every day, and the before-and-after is obvious to anyone who has watched the retyping.
Shipment Visibility
Every hour your team spends hunting across systems is an hour nobody is selling or solving. This is the one your customer service team will notice first.
Knowledge & SOP Search
Read-only, answers from your own SOPs and tariffs, touches no load and no payment. The easiest first yes you will get from anyone who has to approve it.
Bring us the exception that eats the most hours and we will tell you straight whether an agent is the right answer for it.
Fair Enough. The Longer Answers.
We already bought a visibility tool and nothing changed. Why is this different?
Because most visibility products give you another screen. The data lands somewhere new, and your team still opens the TMS, the portal and the text thread to answer one question, because the new screen does not have all three.
An agent works the other way round. It asks your systems the question a person would have asked, assembles the answer, and says where each part came from. The measurable difference is not data coverage, it is how many places somebody has to look. That is what the 90% at Landstar is counting.
Our documents are terrible. Handwritten, faxed, photographed on a phone.
That is the normal case, not the hard case, and it is why the accuracy bar gets set on your paperwork rather than on a clean sample. A vendor demo that works on a crisp PDF tells you nothing about a crumpled BOL photographed in a truck stop at night.
What matters is not average accuracy but knowing which documents the agent should not attempt. It flags what it cannot read and routes it, with everything it did manage to extract already attached, so the person finishing the job is not starting from scratch.
What happens at peak, or when a partner system goes down?
The agent degrades in a way people can see. It reports what it could not reach, works from what it has, and hands off cleanly rather than guessing. A silent failure at peak is worse than no agent, because dispatch stops trusting it and quietly goes back to phone calls, and you never get them back.
Cost behaves the same way. Per-workflow spend is metered and capped, so a volume spike does not turn into a surprise bill in the month you were busiest.
Will these work with our TMS and our EDI?
Yes, and the integration is usually the honest majority of the work. We have built against TMS platforms, telematics feeds, EDI, load boards, carrier portals with no real API, and the receivers whose process is still a scheduled file. The unglamorous plumbing is most of what a working agent actually is.
Read-only first, always. The agent reads your systems and writes to its own record. Writing back into the TMS is a separate conversation and it comes later, with gates.
Status writeback is the exception worth planning for early. If you owe a shipper 214s on a scorecard, an agent that knows a load is late and cannot update the TMS has only moved the work: you stopped searching and you are still keying. So we scope the 214 path in the first conversation rather than discovering it in month three.
Can an agent tender a load or release a payment on its own?
Not unless you decide it should, and we would start by saying no. Tendering, releasing funds and committing to an appointment are money and relationship decisions. The agent prepares them, attaches the evidence and puts them in front of a person.
The value is in everything before that: the reading, the chasing, the matching and the assembling. That is where the hours go, and it carries none of the risk of an agent that spends money by itself.
Should we build these, or buy a platform?
Buy where a product already owns the category and the problem is generic. HappyRobot and FleetWorks for voice check calls, project44 for multi-carrier visibility data. If your problem is the same problem every carrier has, someone has already productised it and buying beats building.
Build where the work is shaped like your operation: the seams between your TMS and everything else, the documents in your formats, the settlement rules your customers negotiated, the SOPs only your dispatchers know. No vendor is going to build those for one carrier.
We will say buy when buying is right, including when the answer is that you do not need us for this one.
We pay an offshore BPO to key documents. Is this actually cheaper?
Probably not per document, and we would rather say so than lose the argument on your second call. A BPO seat is cheap, and it is elastic: you add seats for peak and stop paying in January. Software is a fixed cost that does not shrink when volume does.
So the comparison that matters is not price per document. It is cycle time and error cost. A POD you need today for a detention claim does not arrive on a twelve-hour timezone lag. An extraction that flags what it could not read does not quietly key the wrong rate. And the agent does not need retraining every January when the team turns over.
The number that decides it is what happens to the residual. Automate most of a queue and the exceptions remain, and if they land back at the BPO you have shrunk a contract that probably has a volume minimum, or they land on domestic staff at a much higher loaded cost per exception. We size that with you during the diagnostic, because it is the number that determines whether this is worth doing at all.
What actually changes on the P&L?
One of two things, and you should decide which before you buy, because they lead to different builds. Either the same team moves more freight, in which case the number to watch is loads per person per day and whether margin per load holds while it rises. Or the team gets smaller, in which case you need a role-level plan and an honest view of contract minimums and severance.
We will not promise you a headcount number on a website, and be careful of anyone who does before they have seen your volumes. What we will do in the diagnostic is measure the current cost of the workflow, model both paths on your actual numbers, and give you a figure your CFO can argue with.
Hours given back are not money until somebody decides what the hours are for. That decision is yours, and it is worth making before the build rather than after.
Do we have to take all ten?
No, and nobody ever has. You take one workflow, it either gives hours back or it does not. The second conversation is much easier because the integration and the controls are already in place.
What does it cost to find out if one of these is worth doing?
A fixed-fee diagnostic: two weeks, a clear verdict on the workflow, the integration surface mapped, and a build estimate you can take to your budget holder. Delivered as the Lumynate ROI Audit, from $30K, with half the fee crediting against the build.
For the build itself, first builds are fixed scope with a fixed ship date and typically land in the low-to-mid six figures, driven almost entirely by how many systems the agent has to touch.
If the verdict is that an agent is the wrong answer, you get that in writing, and the finding is usually worth more than the fee.
Every agent above is built on Lumynate, our system for building and operating agents: the method, the reusable component library, the guardrails, and the team that keeps it running after go-live.
Bring Us the Exception That Eats the Most Hours
Thirty minutes, one workflow, and an honest answer about whether an agent belongs anywhere near it.
Or start with the two-week diagnostic if you already have a pilot that stalled.