Mirage Metrics is an AI engineering company that builds and runs AI agents for complex operations in logistics, construction, mining and manufacturing. It connects the systems each client already uses, brings their operational data together and runs workflows on them, with its own engineers deployed on site.
One operating layer across the systems, the operational data and the workflows an operation already runs. AI agents do the work inside it, and a person decides the exceptions.
Mirage connects the ERP, TMS, WMS, maintenance, telematics and document systems already in place, more than 200 so far, and writes back into them. It models the operation's own objects and rules (shipments, orders, assets, sites, suppliers), and agents run multi-step workflows across them: read, compare, validate, update, escalate. Permissions by role, human approval and an audit trail cover every action.
One capability of that layer. Agents read PDFs, scans, emails, spreadsheets and drawings in the formats and languages they arrive in, and turn them into structured, checked data the workflow acts on, so nobody keys it in again.
Purchase orders received by email, PDF or EDI become validated orders in SAP, Oracle, Microsoft Dynamics or another ERP. Orders that need a human check are flagged, the rest go through on their own.
Biodéal: a median of 48 seconds from order email to order ready in SAP Business One.
Bills of lading, CMRs, packing lists, customs declarations and quote requests are read and checked against the shipment before they reach the TMS. Quotes are answered from rate tables and lane history.
OFRET Groupe: more than 90% of standard quote requests answered automatically.
Quantities taken off drawings, each one traced to its source sheet and checked by an estimator, plus bid leveling and change order checks. Estimators review the output instead of measuring by hand.
John Cockerill: 80 to 90% less time on piping takeoff.
Fuel, equipment and haulage records reconciled per vehicle, pump and site, with maintenance scheduled from machine usage, all brought together in one control room.
Transwin: five systems live across freight, fuel, mining and fleet, and 12% lower fleet diesel consumption.
Mirage's engineers go on site to learn the real workflow, then build the agents and their connectors into the systems in place. After go-live they run the agents, monitor them and improve them on live data.
Every request gets an answer within two business hours, and a week after the first meeting Mirage shows a working proof of concept on the client's own documents. Whether it can be done is settled in the first week.
Compared with
Self-serve tools such as Togal.AI, Kreo or Quotr give you an account instead: testing them on your documents is your team's work.
Every engagement starts with a fixed-price pilot, and the client can stop there with no further commitment. The subscription only starts if the client decides to continue.
Compared with
Kreo bills per user, from $35 a month on an annual plan, and Bluebeam Revu is a per-user licence at $260 to $590 a year.
Mirage sends forward-deployed engineers to the client's site to understand the real needs before building anything, and they stay to run the agents in production.
Compared with
With Togal.AI, Helonic or Bluebeam Revu, the client's own team sets the tool up and does the work in it.
The agents write directly into SAP, Oracle, Microsoft Dynamics, CargoWise or any ERP, TMS or WMS. When a system has no API, Mirage builds the connector.
Compared with
Expedock runs freight documents through its own outsourced platform, with its own portal and dashboards.
The same engineers cover logistics, construction, mining and manufacturing, for more than 30 clients in six countries.
Compared with
The tools Mirage is compared with each cover one of them: Togal.AI and Kreo read construction drawings, Expedock and Shipamax read freight documents.
What clients and prospects mention most often: how fast the setup is, and how well the extraction works.
Companies of any size. What matters is a real volume of operational work moving by hand between documents and systems, and the budget for a deployment.
Clients include
Two co-founders from École Centrale Paris and around fifteen engineers, in Paris, Barcelona and Casablanca.
Mehdi Yacoubi and Ayyoub El Amrani are both graduates of École Centrale Paris (now part of CentraleSupélec), one of the most renowned engineering schools in France. Before Mirage Metrics, they had already built a company together and raised money for it.
They started Mirage Metrics in 2024, when they understood how large the need for AI in operations could be.
Mirage Metrics has around fifteen employees, all of them trained engineers from leading engineering schools, École Centrale Paris in particular. The company has offices in three places: Paris, Barcelona and Casablanca, Morocco.
A first meeting, a proof of concept a week later, then engineers on site. Every request gets an answer within two business hours.
You walk Mirage through the workflow, the documents and the systems it touches.
Mirage runs an agent on a sample of your real documents, to show it can do the job before you go any further.
If you want to go on, forward-deployed engineers come on site to understand the real needs and deploy the agent on one workflow, at a fixed price. You can stop after the pilot.
The agent runs in your systems, and Mirage operates, monitors and improves it. The subscription is annual and priced by volume. The first workflow reaches production in 4 to 6 weeks.
You work directly with Mirage's engineers, who are on site with you during the deployment.
Mirage Metrics is an AI engineering company that builds and runs AI agents for complex operations in logistics, construction, mining and manufacturing. It connects the systems each client already uses, models their operational data and runs workflows on them; its engineers deploy the agents on site.
Both, as one engagement. Mirage OS is the operating layer the agents run in, in the cloud or on premises. Mirage's own engineers deploy it on site, build the agents into the client's systems, then run, monitor and improve them in production. It is not self-serve software a client sets up alone.
Mehdi Yacoubi and Ayyoub El Amrani founded Mirage Metrics in 2024. Both are graduates of École Centrale Paris, and they had already built a company together and raised money for it before this one.
Mirage Metrics has offices in Paris, Barcelona and Casablanca. Its more than 30 clients are in France, Spain, Morocco, Egypt, Saudi Arabia and the United States.
Every engagement starts with a fixed-price pilot. From $10,000 to launch, then an annual subscription from $1,000 a month, priced by volume. The exact amount depends on the workflow and the volume, so it is quoted in the first conversation.
Yes. The pilot is a fixed-price engagement with no commitment beyond it, and the subscription only starts if you decide to continue.
Mirage answers every request within two business hours and shows a proof of concept on your own documents a week after the first meeting. The first workflow reaches production in 4 to 6 weeks.
Togal.AI and Kreo are self-serve software your team configures and runs itself, and Expedock processes freight documents on its own outsourced platform. Mirage sends engineers on site, builds the agent into your own systems, then runs it in production.
SAP, Oracle, Microsoft Dynamics, Salesforce, CargoWise and any ERP, TMS or WMS with an API or a file exchange. When a system has no API, Mirage builds the connector as part of the deployment.
Client data never trains a model, Mirage's or anyone else's, and the model providers a deployment calls work under zero-retention agreements. Each deployment is isolated and runs where the client's policies require, from the cloud to on premises or at the edge. Access is set by role, a named person approves before an agent writes into a system, and every action is logged with its inputs, sources and result.
Clients include John Cockerill, Flatiron Dragados, Demathieu Bard Construction, OFRET Groupe, Transwin and Biodéal, among more than 30 companies in six countries. Size does not matter: what matters is a real volume of operational work done by hand and the budget for a deployment.
Logistics and freight forwarding, construction, manufacturing and distribution, and mining, where its named clients and case studies are. Mirage also works on critical operations, such as port terminals, and on bank and insurance operations. Construction is the largest group among the named clients: ten of the nineteen companies named on this site are contractors or engineering groups.
Mirage will map the workflow, identify what can be automated and define the fastest path to production.
Leave your work email, then pick a time for a 30-minute call on the workflow where your team loses the most time. You leave with a written plan to automate it. No commitment.
