5 workflows FourKites customers are automating with Loft:
A carrier hasn’t provided an update on a shipment due at 2 p.m. An exception fired an hour ago and is sitting in a queue. A customer emailed asking for a delivery window, and answering it means checking three different systems. It’s a Tuesday morning, and already the day is full.
Each of those situations requires someone to take multiple steps to resolve. The carrier needs a phone call. The exception needs someone to look it up, figure out what changed, and decide what to do next. The customer’s email means pulling ETA data from one system, cross-referencing it in another, and writing back. Instead of work that requires deep thinking and decision-making, people are coordinating necessary but manual tasks. Most transportation teams spend their days this way.
AI agents for supply chain change what happens after a problem surfaces. For anyone unfamiliar with the term, an AI agent is software that monitors a situation, follows a set of rules defined by the team, and executes the next step on its own. If an ETA shifts and an appointment needs to be rescheduled, the agent reschedules it. If a carrier goes silent on a shipment that’s running late, the agent sends the follow-up. The system does the work, operating within whatever guardrails the team puts in place.
The coordination work that eats most of the day can be handled by software. People get to focus on the problems that need a human brain.
FourKites customers are using Loft, an AI orchestration platform, to build automated workflows tailored to how their teams operate. The emphasis here is on tailored. Teams describe their own processes, and Loft builds working agents around them. Sophie, FourKites’ AI developer agent inside Loft, is what makes that possible. You describe a workflow in plain English, and Sophie translates it into a production-ready agent.
FourKites has pre-built agents that prove the model works at scale, but the custom workflows that teams are creating for their own operations are the bigger story. Here are five examples.
Automated carrier follow-up on at-risk shipments. When a carrier hasn’t checked in on a shipment that’s approaching its delivery window, an agent picks up the follow-up. It contacts the carrier, logs the response, and escalates if the update doesn’t come back in time. The person who used to spend twenty minutes chasing that information now gets a notification with the answer.
Appointment rescheduling when ETAs shift. An inbound shipment’s ETA changes, which means the dock appointment no longer lines up. Normally someone catches this in a report, calls the warehouse, and manually reschedules. An agent built through Loft detects the ETA change and reschedules the appointment automatically, then notifies the relevant teams.
Automated, proactive customer delivery updates. Customers ask for delivery status when they don’t have insight into what’s happening. An agent monitors shipments tied to customer orders and anticipates when a customer will want an update, pushing it before they have to ask. When customers do still reach out with questions, the agent can field those inbound inquiries too. The person who used to spend their morning answering status emails is now handling the situations that need judgment.
Document collection and freight audit. After delivery, someone has to collect bills of lading, proof of delivery, and other documents, then match them against the original shipment record for audit. An agent handles the collection, flags discrepancies, and routes exceptions to the right person. The routine matching happens in the background.
Custom workflows built to the team’s SOPs. Loft lets teams describe any process to Sophie and get a working agent for it. If a team has a specific escalation path, or a set of rules for how to handle a particular type of exception, or a process that currently lives in someone’s inbox or in a shared spreadsheet, Sophie can build an agent around it. The process doesn’t have to fit a template.
When a team builds a workflow through Loft, the individual procedures that make up that workflow are called Agent Operating Procedures, or AOPs. They’re written in plain English based on the team’s existing processes, and they’re reusable. Once you’ve built an AOP, it becomes a building block for the next workflow. Your second custom agent deploys faster than your first. Your tenth deploys faster than your fifth.
AOPs also record the reasoning behind each action an agent takes. When a shipment gets rescheduled or a carrier follow-up gets escalated, the system logs why it happened and what rules were applied. For teams that are cautious about handing work to software — and they should be — that visibility matters. You can see what the agent did and why, and you can adjust the procedure if the logic needs to change.
The traditional answer to “how long does it take to automate a workflow” has been months. IT gets involved, integrations get scoped, someone owns the maintenance forever. With Loft and Sophie, teams describe what they need and the workflow is production-ready in days.
That speed matters more than it might seem. According to McKinsey, major supply chain disruptions now occur roughly every 3.7 years. Teams that take six months to stand up a new workflow are building for conditions that may have already changed by the time it goes live. The ability to spin up a new agent in days means the team can respond to what’s happening now, not what was happening when the project kicked off.
On April 27 at 11 a.m. CT, FourKites is running Loft Live, a session where an audience member submitted a real workflow and Shana Wray builds it on screen during the event. A real workflow, from description to working agent, built live.
If you want to see what this looks like in practice, register here.
The workflow that’s currently living in someone’s head as tribal knowledge, or buried in a shared spreadsheet that three people know how to use — that’s the kind of thing Loft was built for. And on April 27, you can watch it happen.
