How OFRET Groupe Automated Quoting Across Two Business Lines with Mirage
OFRET Groupe runs two distinct operations under one roof — OFRET for full truckload transport, SWILL for groupage. Mirage deployed a tailored CargoScribe Quoting Agent for each, both connected to the group's shared TMS.
About OFRET Groupe
OFRET Groupe is a freight forwarding group built around two entities with distinct jobs. OFRET handles full truckload transport — complete shipments moved directly from origin to destination. SWILL runs groupage — consolidating multiple clients' smaller shipments into shared loads. Same group, same TMS, two different pricing logics and two different sets of carriers to manage.
Both entities live with the same underlying pain: a constant stream of quote requests to price by hand, and a pile of freight documents to read and re-key into the system for every shipment. Done manually, that work does not scale — and it does not scale the same way for full truckload as it does for groupage.
Two entities, two environments
Because OFRET and SWILL quote differently — different carrier networks, different document types, different rules for what makes a request "standard" — Mirage did not deploy one generic agent across the group. CargoScribe was set up as two separate, purpose-built environments, one tuned to OFRET's full truckload workflow and one tuned to SWILL's groupage workflow.
Both environments plug into the same group TMS and share the same underlying platform, so improvements made for one — a new verification rule, a better extraction pattern — can be adapted for the other without starting from zero.
Where it started
OFRET Groupe already ran its operations on Logitude World, its TMS. Rather than add another software layer on top, Mirage connected directly to the Logitude World API and built the automation inside the system both entities already used.
Mirage's forward-deployed engineers worked on-site with OFRET's and SWILL's own documents and workflows, not a generic template. That embedded approach — building against the real carriers, the real document formats, the real exceptions each team runs into — is what made it possible to stand up two tailored environments instead of one compromise.
The Quoting Agent
At the center of both environments is a CargoScribe Quoting Agent. When a client emails a request to move a shipment, the agent reads the RFQ, extracts the shipment data it needs to price it — origin, destination, weight, volume, incoterms, deadlines — and checks it against the entity's own carrier and subcontractor network to prepare a quote.
For the large majority of requests, the standard cases, the agent prepares the response automatically, cutting the time between an incoming RFQ and a client getting a quote from hours to minutes. Everything else — the requests that do not fit the standard pattern — is routed for review rather than guessed at.
From documents to full RFQ automation
The deployment did not arrive fully formed — it grew in stages. Mirage started with document automation: reading freight documents across the TMS and turning them into structured, usable data. Once that extraction was reliable, it became the foundation for RFQ automation — reading incoming quote requests and preparing priced responses directly.
As trust in the agent's output grew, Mirage layered in a verification stage: rules that check a quote or an extraction against what "normal" looks like for that entity, and an exception layer that catches anything that fails those checks and notifies an operator instead of letting it through. Each stage was built on the one before it — an automation stack that keeps evolving rather than a one-time delivery.
Human in the loop
The agent handles the routine. People handle the judgment calls. Specific triggers send a case to a human operator before anything goes out:
› Missing data on a shipment document
› Low-confidence extraction the agent isn't sure about
› Pricing that falls outside an entity's standard rate patterns
› Requests that don't match either entity's usual shipment profile
Because Mirage's engineers were embedded with OFRET's and SWILL's teams while building this, the exception rules reflect real cases the teams had actually run into — not a generic guess at what might go wrong. That forward-deployed relationship stayed in place after launch, so the rules keep getting refined as new edge cases surface.
What changed
Neither OFRET's nor SWILL's teams re-key documents or price standard RFQs by hand anymore. The routine quoting that used to consume the day now runs automatically inside the group's own TMS, tailored to each entity's way of working, and people focus on the shipments that actually need a decision. What started as one document workflow has grown into an evolving automation stack — and made Mirage a recurring AI partner OFRET Groupe comes back to for new deployments.
Where it goes next
› Customs documentation
› Shipment tracking
› Automated client communication
› Collections
“Mirage did not sell us a product to install. They connected to the tools we already use and built a quoting agent for each side of our business, then kept improving it as we found new edge cases. Each time we had a new need, they built it. They have become our AI partner.”