A container terminal at night, seen from the air.
enzwa lab  ·  concept

Meridian

A layer of autonomous agents over every node of a supply chain. It watches five systems that never speak to each other, and works the problem before anyone picks up the phone.

A tool for showing what an automated supply chain does the moment something breaks.

The problem

Orders, stock, shipping and suppliers usually sit in four systems that never speak to each other. A missed delivery gets noticed days later, by which time the cheap options have gone.

What it does

Meridian puts a layer of software agents across all four, so a break is caught as it happens and rerouted while a person is still reading the alert. Break the chain yourself and watch it repair.

the two days nobody counts
40 hours

Published research puts the average time to detect a supply-chain disruption at eight point seven hours, and a full read on what it means at more than forty. That is nearly two days in which the chain keeps running on a plan that stopped being true.

Seventy-two per cent of manufacturers learn about a shortage once the delay is already unavoidable. By then the moves that remain are the expensive ones: air freight, penalty clauses, and a call to the customer.

a chain with no agent layer nobody knows working out what it means 8.7 h · detected 40 h · first real decision the gap, to scale the same chain, with Meridian 2 h 40 · re-planned and repaired seen at four seconds 0 h8 h 16 h24 h 32 h40 h 48 h
swipe the chart →
time before anyone knows time spent working out the impact seen, re-planned and repaired
A container terminal under heavy rain at night.
one closed port  ·  three stopped lanes  ·  nothing in the chain has been told
what it watches

Five systems that never speak to each other.

Orders, inventory, transit, suppliers and finance each live in their own tool, with their own owner and their own refresh. A person joins them up by opening five tabs and remembering what the sixth one said.

Meridian puts an agent on every one of them, so the chain stops being five dashboards and becomes a single living graph: one place where a signal in transit is already a number in finance.

ordersinventory transitsuppliers finance one agent each one living graph nodes · lead times dependencies to the orchestrator
swipe the diagram →
how the agents respond

A control layer that reasons.

A dashboard tells you something broke. Three principles govern what Meridian does about it, and each one is a property of the graph rather than a rule somebody wrote down.

01

Branching

One signal opens a decision tree. The orchestrator prices every branch at the same time and commits to the one that resolves the case.

pricedpriced taken
02

Redundancy

Critical paths are mirrored. A warehouse empties and the second one serves; a supplier defaults and the backup takes the line. Failing over is a routine move.

primarymirror
03

Dependency

A parcel cannot be traced without the tracking reference, and a claim cannot be filed without the carrier’s. The graph holds who waits on whom.

tracking reftraceclaim waits on
A motorway interchange from above at night, dense with light trails.
every option priced at once  ·  one committed to
one closure, worked end to end

What the two hours look like.

A transhipment port closes on a Monday morning and takes three lanes with it. Nothing below is a person making a call.

T+ 00:00
The port closes. Three lanes stop, and everything behind them is now late.
T+ 00:00:04
Transit has the closure, orders has the exposure and finance has the number. Four seconds.
T+ 00:01
The orchestrator opens the tree and prices every branch at the same time: alternate carriers, alternate ports, alternate sequencing.
T+ 00:06
The mirrored distribution centre comes up and the backup supplier takes the line.
T+ 00:14
The recovery re-sequences around what gates what, so the claim waits for the carrier reference instead of failing on it.
T+ 02:40
New lanes carrying, the mirrored centre serving two markets, the claim filed and the exposure priced.
T+ 40:00
The hour at which a chain with no agent layer typically gets its first full read on what the closure meant.

Detection and impact timings, revenue loss and daily cost are published 2025 and 2026 supply-chain research. The closure above is illustrative, and every figure inside it is synthetic.

A vast distribution warehouse at night with one aisle lit.
two hours and forty minutes  ·  the chain whole again
what the gap is worth

Two days, priced.

An agent layer will not stop the typhoon. What it removes is the forty hours between the typhoon and the first good decision, which is where most of the money goes.

$1.5m
the average cost of a single day of disruption
2.4%
of annual revenue, lost to disruption on average
44.5%
of organisations lose three to four per cent of revenue
the model is live

Give it your own chain.

Three questions about what you move and where, then the mesh runs your case: the live supply graph on one side, the agents reasoning branch by branch on the other. Break something and watch it repair.

Open Meridian →