01 / 11
AI Operations Playbook

From 5 Dispatchers
to 1 Super-Dispatcher

An army of AI agents takes over 80% of the routine. One senior operator owns the hard 20%. Throughput grows 3–5×.

80%
Routine automated
5 → 1
Headcount
3–5×
Fleet throughput
0 vehicles tracked now
0 AI calls today
Logistics AI command center
Live · 0 vehicles tracked
Why this role

The dispatcher job is three clean loops — perfect for AI.

Data Monitoring

GPS, statuses, ETAs, telematics signals streamed 24/7.

Communication

Calls to drivers and clients the moment something slips.

Rule-Based Decisions

If late > 30 min → notify client. If breakdown → call tow.

1
Signal
2
Detect
3
Decide
4
Act
Step 1 — Capture

Digitize the silent experience.

Every dispatcher action becomes a row on a "Golden Timeline" — the training set that teaches the AI what humans actually do.

GPS Telemetry
Location, speed, idle, geofence events
Screen / OCR
TMS, ERP, CRM tabs the dispatcher reads
Call Transcripts
Driver & client conversations, STT-labeled
CRM Updates
Status changes, notes, exceptions logged
14:02:11 GPS truck #A-417 stopped on M-4, 12 km from delivery
14:02:43 OCR dispatcher opens TMS → order #88231
14:03:05 CALL "Sergey, what happened?" → "Flat tire, 40 min"
14:04:18 CRM status: DELAYED · ETA +45m · client notified
→ Pattern learned: stop > 5min + off-route → call driver → update ETA → notify client
Step 2 — Build

Three agents. One operation.

Monitoring agent

The Eye

Watches every truck. Flags deviation, idle, off-route, speeding the instant it happens.

Telematics APIGeofencesAnomaly rules
Communication agent

The Voice

Calls drivers in their language. Reads back the answer and posts it into CRM.

TwilioElevenLabs TTSWhisper STT
Decision agent

The Brain

Re-plans routes, picks the tow company, escalates the right case to a human.

OR-ToolsLLM routerPolicy engine
Signal
The Eye detects
The Brain decides
The Voice acts
Human approves edge cases
The Eye, live · click any truck

412 trucks. One screen.

#D-501
On schedule
At risk
Incident
Click any dot for details
Truck
#D-501
E-95 · Sofia → Bucharest
Incident
ETA
—
Delay
+95 min
Reason
Breakdown — flat tire
Dispatcher actions
  • AI: tow dispatched (ETA 35 min)
  • AI: backup truck assigned
  • Human: key-account notified
Step 2 — Rules in action

Pick a trigger. See the agent reply.

Each rule fires the right agent automatically — The Voice for calls, The Brain for orchestration. Click a trigger to preview the generated action, copy the script, or verify rule compliance.

Signal
ETA slipped by 42 min on order #88712 (key client SLA = 30 min)
Agent
The Voice
→ Client · LogiStar Co.
Outbound voice call · ElevenLabs TTS · ~22s
Generated call script · preview
“Hello, this is the LogiCorp automated dispatch.”
“Your shipment #88712 is 42 minutes behind schedule.”
“New estimated arrival: 17:48. We have rerouted the truck via A-104.”
“Reply 1 to confirm, 2 to talk to a human operator.”
Rule compliance
  • ·
    Mentions order ID
  • ·
    States new ETA
  • ·
    Offers human escalation
  • ·
    Identifies the company
Step 4 — Impact

The numbers that change.

Response time to deviation
7 min → 8 sec
Fleet utilization
+20 pp in 6 months
500–1k
Trucks per operator
vs 80–120 today
24/7
Coverage
no shifts, no fatigue
100%
Consistency
same SLA every call
−68%
OPEX of dispatch desk
payroll + tooling
Step 3 — Human-in-the-Loop

Triage by color. Humans only on red.

super-dispatcher · live queue
processing
Truck #A-204 · accident on E-95
Awaiting human · key client
Order #88712 · delay 42 min
AI drafted client message — approve?
Order #88715 · ETA shifted +10 min
AI auto-notified client ✓
Truck #B-118 · refuel completed
AI logged to CRM ✓
Shadow mode
Week 1–2: AI predicts, humans act. Compare every decision.
Auto-handled today
1,284 / 1,310
98% routine cases closed without a human
MVP · 6 weeks

From zero to autopilot.

W1

Wire up telematics

Connect GPS/TMS API, start screen & audio capture.

W2

Train The Voice

Twilio + ElevenLabs + Whisper, basic call flows.

W3

Build The Eye

Geofences, anomaly rules, deviation alerts.

W4

Shadow Mode

AI predicts every move, humans still act. Measure agreement.

W5

Auto-handle routine

Green-zone cases close without a human.

W6

Full rollout

Only red zone reaches the operator. Throughput 3–5×.

Why it pays

1 senior salary replaces 5.

Before
5 mid-level dispatchers · ~80 trucks each
After
+
1 senior operator + army of agents · 500–1,000 trucks
Net effect
3–5× throughput · −68% cost
Summary

The dispatcher of the future is a team of agents.

You don't hire more people. You hire one great operator and give them an army.

Agents handle 80%
Routine calls, ETA updates, status sync — fully autonomous.
Humans own 20%
Edge cases, key clients, escalations — where judgment matters.
3–5× throughput
Same operator now runs 500–1,000 trucks, 24/7, no fatigue.
The Eye
The Brain
The Voice
Human (hard 20%)
AI Super-Dispatcher · Logistics Operations Playbook