HamiltonTry the live demo
AI operations for small delivery fleets

Your deliveries run themselves. You stay in charge.

Hamilton’s agents plan routes, fix them when the day breaks, answer customers and check every proof of delivery. Anything that touches money or people waits for your OK.

Try the live demoSee how it works
  • No sign-up
  • Runs in your browser
  • 2 minutes
Today, 10:42Maria’s Flowers & Gifts2 at risk
JunFlat tire
Carlo
Ana
Rico
✓ 1039✓ 1041104210431044✓ 1045✓ 1046✓ 10471048104910501051
Dispatch agent · HAM-417
Move 3 of Jun’s stops to Ana and Rico?
Jun has a flat (about 1 hour). If you do nothing, #1042 and #1049 arrive 40–85 min late. With this change all 3 arrive on time.
Held for you: 3 stops is over your 1-stop auto limit.

↑ This is live. Approve the fix and watch the routes change.

Built for fleets of 2–50 driversFloristsBakeriesDistributorsPharmaciesLocal couriersGift shops
The problem

In a small fleet, the owner is the dispatcher, the help desk and the claims team.

Routes get rebuilt from memory, “where’s my order?” arrives all day, and a disputed delivery turns into an argument about a blurry photo. Enterprise tools are too heavy and too expensive. Tracking apps only track.

8%
of last-mile deliveries fail on the first attempt
$17
is the cost of each failed delivery
94%
of customers blame the shop, not the courier
1.7 h
a day goes to manual dispatch and follow-ups

Sources: OneRail via FreightWaves; Microlise survey of 250 logistics managers. Full notes in the research doc.

Three agents, one team

They do the legwork. You make the calls.

Plans the day. Re-plans it when the day breaks.

Builds tomorrow’s routes from your drivers, vans, history and address book. When a tire goes flat or a road closes, it finds the fix in seconds and shows you the before and after.

  • Morning plan in one click
  • Live re-planning with arrival times you can trust
  • Holds a slot open for same-day orders
See it in the demo →
StopIf you do nothingIf you approve
#1042 Mark Tan13:40 · late12:15 · Ana
#1050 Nicole Ramos14:0512:40 · Ana
#1049 Sam Ocampo13:55 · late11:05 · Rico
Autonomy you can dial in

Agents propose. Your policies decide.

Every action goes through rules you set in plain language. Start with “Ask me” for everything, then let Hamilton earn more autonomy as you approve its work. Try it:

1 · An agent proposes
2 · Your policy for reassigning stops
Auto if moving
1
stop or fewer
3 · What happens
Becomes a ticket for you
Opened as a HAM ticket with the before and after. Nothing changes until you approve.
policy.check()3 stops is above the 1-stop auto limit.

This box runs the same policy code as the demo. Agents can’t talk their way past it: the check is in code, not in a prompt.

How it works

From a flat tire to a fixed route in under a minute.

  1. 1
    Something happens
    A driver taps “Flat tire”, a customer replies, an order lands from Shopify or WhatsApp.
  2. 2
    Triage sorts it
    Fast, low-cost checks: what is this about, is it urgent, does it need a person?
  3. 3
    An agent drafts the fix
    A reasoning model with tools re-plans routes, writes the reply or builds the evidence pack.
  4. 4
    Your policy decides
    Auto runs and is logged. Ask me becomes a ticket. Never is blocked. Every step is in the audit log.
Every action is logged
Input, decision and side effects, with Undo.
AI output is always labelled
You always know what a person did and what an agent did.
Money is never automated
Refunds and credits always go to a person.
The 2-minute demo

Live a Monday at Maria’s Flowers & Gifts.

Four drivers, thirteen orders, Metro Manila. At 10:31 Jun gets a flat tire. Be the driver, be the owner, and see what the customer sees.

Start the demoOpen the driver app
  1. 10:30
    A calm morning
    Everyone on time.
    Open →
  2. 10:31
    Jun reports a flat tire
    One tap in the driver app.
    Open →
  3. 10:33
    The agent proposes a fix
    Held for you as HAM-417.
    Open →
  4. 10:44
    You approve
    Routes, ETAs and customers update.
    Open →
Roadmap

Autonomy is earned, not switched on.

Now
Proof of concept
Admin console, driver app and customer view on one scripted day. Policy engine in code.
0–3 months
Design-partner pilot
3–5 fleets in one city. Shopify, WhatsApp and tracking-tool integrations. Agents start in “Ask me”.
3–6 months
Earned autonomy
Approval rates suggest which rules can move from “Ask me” to “Auto”. Driver voice notes.
6–12 months
Platform
Weather and traffic in tomorrow’s plan, multi-depot fleets, and an MCP server for owners’ own assistants.

Watch a morning go wrong, and get fixed, in two minutes.

No sign-up. No API keys. Nothing to install.

Try the live demo