Free MCP server · No API key

Route optimization for your AI agent.

Paste a list of addresses into Claude, ChatGPT, or any MCP-enabled agent and get back the optimal visiting order — with ETAs, leg distances, and total route stats. Powered by the same Routific engine behind 191 million+ optimized deliveries.

https://mcp.routific.com/mcp
Free — no signup 8 stops per route Addresses geocoded automatically
How it works

From address list to optimized route in one message

The server exposes a single tool, optimize_route, that your AI agent calls whenever you ask it to plan a multi-stop trip.

1

Connect

Add mcp.routific.com/mcp as a connector in Claude, ChatGPT, or any MCP client. No account, no API key, nothing to configure.

2

Paste your stops

Share up to 8 addresses in plain text. Routific geocodes them automatically — just include the city and region.

3

Drive the best route

Claude returns the optimal visiting order with an ETA and driving distance for every stop, plus total time and distance.

Live example

A real conversation, a real route

This is actual output from the server, shown in Claude — six stops around Vancouver, optimized in about two seconds. It works the same in ChatGPT or any MCP-enabled agent.

You → Claude
Plan my delivery run for tomorrow, leaving the office at 8:30am and ending back here: • 800 Robson St, Vancouver • 2305 W 7th Ave, Vancouver • 5300 No. 3 Rd, Richmond • 6200 University Blvd, Vancouver • 3211 Grant McConachie Way, Richmond Office: 455 Granville St, Vancouver
optimize_route · Routific MCP
Optimized route · departs 08:30
  1. S
    455 Granville St — start
    departs 08:30
  2. 1
    800 Robson St
    eta 08:32 · 0.6 km
  3. 2
    3211 Grant McConachie Way (YVR)
    eta 09:03 · 13.3 km
  4. 3
    5300 No. 3 Rd, Richmond
    eta 09:18 · 6.4 km
  5. 4
    6200 University Blvd (UBC)
    eta 09:54 · 17.4 km
  6. 5
    2305 W 7th Ave
    eta 10:18 · 8.6 km
  7. S
    455 Granville St — return
    eta 10:36 · 4.7 km
101 min
total travel
50.9 km
total distance
7 stops
sequenced
Setup

Connect in under a minute

The server speaks Streamable HTTP — the standard MCP transport. Pick your client:

  1. Open Settings → Connectors in Claude (web or desktop).
  2. Click Add custom connector.
  3. Name it Routific and paste the URL below — no authentication needed.
  4. Ask Claude to optimize a route. That’s it.
https://mcp.routific.com/mcp
  1. In ChatGPT, open Settings → Apps & Connectors and enable Developer mode (under Advanced settings; available on paid plans).
  2. Click Create, name it Routific, paste the URL below, and set authentication to None.
  3. In a chat, enable the connector from the + menu, then ask for an optimized route.
https://mcp.routific.com/mcp
  1. Run this in your terminal:
claude mcp add --transport http routific https://mcp.routific.com/mcp
  1. Endpoint: POST https://mcp.routific.com/mcp (Streamable HTTP, stateless).
  2. No authentication, no session headers required.
  3. Example JSON config for clients that take an mcpServers block:
{
  "mcpServers": {
    "routific": {
      "type": "http",
      "url": "https://mcp.routific.com/mcp"
    }
  }
}
Reference

The optimize_route tool

One tool, four parameters. Routific solves the underlying traveling-salesman problem and returns a schedule with machine-learned ETAs — not just a stop order.

Parameters

NameTypeDescription
stops requiredstring[] · 2–8Addresses to visit, e.g. “800 Robson St, Vancouver, BC”. Include city + region for reliable geocoding.
start_address optionalstringWhere the driver starts. Defaults to the first entry in stops.
return_to_start optionalbooleanEnd the route back at the start. Default false.
departure_time optionalstring24h hh:mm departure used to compute ETAs. Default 09:00.

Response

FieldDescription
route[]Stops in optimal order: address, eta, departure, and distance_from_previous_m.
total_travel_time_minutesDriving + service time for the whole route.
total_distance_kmTotal driving distance.
unservedStops that couldn’t be routed, with the reason (e.g. address not found).
8

Why the 8-stop limit — and why not to split bigger lists

This free demo is capped at 8 stops per request. Splitting a longer list into chunks and optimizing them separately does not produce an optimized route — the chunks are only locally optimal. For unlimited stops, multiple drivers, time windows, and proof of delivery, upgrade to Routific or build on the Engine API.

What’s next

This is just the first stop

We’re actively expanding what AI agents can do with Routific — and your ideas help decide what ships next.

Coming soon

More tools, higher limits

We’re working on additional MCP tools and higher stop limits — letting your agent interact with the full Routific platform: optimizing routes across your whole fleet, visualizing them on a map, dispatching them to drivers, and tracking deliveries as they happen.

Your idea here

Got an MCP use-case?

Building something with AI agents and delivery logistics? We’d love to hear what you’d use a Routific MCP for — your ideas shape the roadmap.

Share your idea →
FAQ

Good questions

Is it really free? What’s the catch?

Yes — it runs on Routific’s public demo API token. The only limit is 8 stops per request. If your routes outgrow that, that’s the catch: we’d love you to try Routific.

Where does my data go?

The addresses and route parameters you provide are forwarded to the Routific Engine API (api.routific.com) to geocode and optimize the route — nothing else is collected, and this server stores nothing. See the Routific privacy policy.

How are the ETAs calculated?

They’re powered by machine learning. The Routific Engine learns from millions of real-world deliveries every month, so its travel-time predictions keep getting more accurate — the same blazing-fast engine behind our route optimization API. Each ETA combines your chosen departure time (default 09:00), ML-predicted driving times between stops, and 5 minutes of service time at each stop.

Do I need exact addresses or coordinates?

Plain addresses work — they’re geocoded automatically. For best results include the city and region (e.g. “800 Robson St, Vancouver, BC, Canada”).

Which MCP clients are supported?

Any client that speaks Streamable HTTP: Claude (web, desktop, mobile), ChatGPT (developer mode), Claude Code, Cursor, and most other MCP-capable tools. The endpoint is stateless, so no session or auth setup is needed.

Need more than 8 stops?

Routific plans routes for entire fleets — unlimited stops, multiple drivers, time windows, live tracking, and proof of delivery. Trusted by 1,000+ delivery businesses.

Explore Routific Engine API docs