Back to Logistics & Freight

Live Demo · Agent 02 · Logistics

A shift with The Exception Scout.

You are the dispatch manager at Meridian Freight. Forty-seven loads are on the board, and an agent has been watching them all night. Clock in and work a full day beside it: what it does, what it refuses to do, and the one decision it hands back to you.

Your clicks run the shift 1 decision only you can make 3 min start to finish
The agent · on shift
the-exception-scout
Agent 02 · watches the board
Watching
The loop it runs
Trigger
a clock. Every 15 minutes, no human needed.
Read
track-and-trace + the order file. Read-only.
Think
three cheap rules first. The model only on candidates.
Act
one post to the channel, or silence. Nothing else.
Your call
the judgment. Stays with a named human. You.
Log
every read, every post, every override. On the record.
The agent's P&L · five lines, written before it touched real work
ScopeWatch every load. Post exceptions to this channel. Nothing else.
OwnerYou. Dispatch manager. One named human, not a committee.
Budget$5.00 a day, enforced in the request path itself. spent today$0.030
ReviewFirst Friday, monthly. Judged on output shipped, like any hire.
TrailOn since day one.
#load-exceptions · the Scout posts here. Humans decide here.
Your desk · Meridian Freight Co.
Wednesday · dispatch desk
06:45 ET
your shift 06:45 to 17:00
the Scout's shift always
47
loads
47
in window
0
watching
0
posted
track-and-trace api · edi · eld
feed idle · next tick on the clock
Wednesday, 06:45

The board is quiet. Clock in.

You will watch three things: the agent and the loop it runs, the channel where it speaks, and your desk with the clock and the live board.

Fair warning: most of the day, nothing happens. That is the point.

Clock in to start
Every read, every post, every decision and every dollar lands here. The audit trail is on before the agent is.

Simulated freight, real anatomy. Meridian Freight is invented and the numbers are illustrative. The loop, the spend cap, the scope boundary and where the human sits are exactly how the production build is wired.

The Architecture

One loop. Any industry.

Everything in that shift was one pattern: a trigger the agent waits on, a read-only look at your systems, cheap rules before expensive thinking, one bounded action, a named human holding the judgment, and a log of all of it. Swap the feed and the post, and the same loop runs any desk we deploy.

Every industry above is an example build, not a case study. Which loop is worth running first on your desk is exactly what the Readiness Map answers.

Prefer the engineer's cut? Open one tick, layer by layer

The Real Numbers

Enough simulations. Get your number.

Every number in that shift is an estimate, because every desk is different. The Readiness Map makes it exact: your workflows, your multipliers, your first agents, with names and owners.

Book the call