Operations software built from engines that work together

Route planning, AI agents, automation, sales outreach and field service, sharing one contact list, one calendar and one set of permissions. Switch on the engines you need.

14-day free trial, no credit card. Or book 30 minutes with us.

Route planner: a day of stops split across four drivers and drawn on a map
AgentsPhone agent answered a delivery question
SalesA lead accepted your LinkedIn invite
FleetStop 14 delivered, photo and signature saved

What a day on it looks like

Four moments from four engines, in one workspace.

  • Fleet

    A day of stops, split across drivers

    Import the stops, press Plan, check the rounds, send them to the driver app.

  • Agents

    A customer call, answered

    The phone agent reads the live route and tells the caller when the driver will arrive.

  • Marketing & Sales

    An invite accepted, a message written for that person

    LinkedIn steps wait for the connection, and every message is drafted for one reader.

  • Field service

    A job closed on site

    The job was dispatched as a route, and the visit is closed from the field with proof of completion.

Why one platform

Many engines, one foundation

Each engine does one job well. They share the same contacts, the same calendar and the same roles, so a stop, a lead and a caller are one record, and work moves between engines without an integration project.

  • One contact record

    The customer on a route is the same contact as in the pipeline and the inbox.

  • One calendar

    Routes, jobs, agent runs and reports scheduled in one place.

  • One permission model

    Each engine declares its permissions; one role model grants them.

  • On per workspace

    Turn engines on for the workspaces that need them.

Surfaces
  • Dashboard
  • Driver app
  • Tracking page
  • Chat widget
  • Phone line
  • API and MCP
Engines, switched on per workspace
Shared foundation
  • Contacts
  • One calendar
  • Roles and permissions
  • Governance
  • Connectors and Drive
  • Audit log

Fleet

Plan a day of stops, split it across drivers, send it to the driver app and follow every vehicle live, up to proof of delivery.

  • Route solver with time windows, capacity and a multi-driver split
  • Driver app for Android and iOS that keeps working offline
  • Proof of delivery: photo, signature, QR or barcode scan
  • Customer tracking page that hides the driver until they are close
Exchanges data with
  • Work orders are dispatched as real routes.
  • Agents read routes and stops as tools.
  • Texts and emails can carry the tracking link.
  • A flow step can add a stop to a route.
  • Trips per day, shifts and exports.
  • Geofences and the yard plan share the map.
More on Fleet
Plan

From a list of stops to proof of delivery

Import the day's stops, let the solver build the rounds and split them across your drivers, then send them to the driver app. Your customers follow their own delivery, and you replay any vehicle's day trip by trip.
  • Respects time windows, appointments and vehicle capacity
  • Splits one day across several drivers, one real route each
  • Driver app for Android and iOS that keeps working offline
The driver app on a stop: navigation, proof of delivery and an offline queue
The customer tracking page: delivery status, and the driver's position shown only once they are close
A tracked vehicle's day: each trip with where it started, where it ended, distance and time
  • Route solver

    Builds rounds that respect windows and capacity, then lets you adjust before sending.

  • Multi-driver split

    One list of stops becomes one route per driver, split on time windows and capacity.

  • Driver app

    Navigation, stop notes and an offline queue that syncs when the signal returns.

  • Proof of delivery

    Photo, signature, QR or barcode scan and the recipient's name, on the stop.

  • Customer tracking page

    Shows the stops ahead and an estimate, and the driver only in the last stops before yours.

  • Trip history and exports

    Every trip with where it started and ended, per day, to CSV or Excel.

Automate

Agents that work inside your systems

Give an agent a model, instructions and tools from the engines you run. It answers chats on your website and calls on your numbers, reads routes, deals and jobs, and asks a person before it sends, browses or starts a flow.
  • Realtime voice agents on your own phone numbers
  • Chat widget you embed with one script tag
  • Bring the model: OpenAI, Anthropic, Google, Mistral, Azure, OpenRouter or a local model via Ollama
A voice agent answering a phone call, looking up a delivery and asking a person to approve a message
  • Tools from every engine

    Agents read and write real records: routes, contacts, deals, jobs.

  • Voice on live calls

    Answers inbound calls in realtime and hands over to a person when needed.

  • Human approval

    Sending email, browsing and running flows wait for a person to approve.

  • Knowledge

    Documents, databases and APIs as sources the agent searches before it answers.

  • Scheduled tasks

    Agents that run on a schedule, on the same calendar as everything else.

  • Governed

    Personal data redacted and every action in the audit log.

Connect

Automations that cross engines without glue code

Flows run on a canvas. A webhook, a schedule or a device starts them, and the steps belong to the engines themselves: create a contact, add a stop, write a message with an agent, send a text.
  • Templates to start from, and every save versioned
  • Signed webhooks and schedules as triggers
  • More than 50 connectors, plus an HTTP step for everything else
A flow that turns a new web order into a contact, a delivery stop, an AI-written confirmation and a text message
  • Canvas builder

    Drag steps onto a canvas, branch on conditions, and see the history of every run.

  • Triggers

    Webhooks, schedules, device events, and email arriving in the inbox.

  • Connectors

    Credentials stored once in an encrypted vault and reused by every step.

Grow

Pipeline and outreach that read like a person wrote them

Find leads, clean the lists you already have, and run sequences on LinkedIn and email that keep a human pace. LinkedIn steps wait until the connection is accepted, and each message is written for the one person it goes to.
  • Daily and weekly limits with human-like timing
  • Connection-aware steps: no message before an accepted invite
  • Drafts you can review before they go out
An outreach sequence across LinkedIn and email with each person's current step and a daily pacing limit
  • Pipelines and deals

    Stages you define, deals you drag, activities and tasks on each.

  • Lead finder

    Search providers such as Apollo, Lusha, Hunter or People Data Labs from one screen.

  • Import and clean

    Stage an old company list, remove duplicates, check emails and classify it before it reaches the CRM.

  • Per-person messages

    Every message written for one reader and checked for length and facts.

  • Browser accounts

    Outreach runs through real browser sessions for the accounts you connect.

  • Deal to invoice

    A won deal becomes an invoice in the Invoicing engine.

Serve

Field work, dispatched like a delivery

A job has a customer, a place, an appointment window and a technician. It goes out through the same route solver as your deliveries, and closes with proof of completion from the field.
  • Board and schedule views for every job
  • Appointment windows the planner respects
  • Agents and flows can create and assign jobs
A board of field jobs with appointment windows, moving from scheduled to done
  • Work orders

    Customer, location, window and assignee on every job.

  • Dispatch through Fleet

    Jobs become real routes for the technician's day.

  • Service deskComing soon

    Internal tasks, bugs and incidents on a board, linked to the conversations they came from.

And the rest of the stack

Inbox and phone, stock, reports, governance, maps and cameras: engines you switch on when the work calls for them.

Measured

Tested on a real delivery operation, not a demo day

We ran the new route solver on nine working days of a Dutch delivery company's actual stops, and on a public benchmark used by the route-planning industry.

18,475 km driven by the new solver
driven by the new solver
vs 19,419 km on the customer's own saved routes
1,230 stops served
stops served
vs 1,155 with our previous planner
70 route crossings
route crossings
vs 97 with our previous planner
~1% from Routific on driving distance
from Routific on driving distance
across 121 problems of Routific's open benchmark

Measured October 2026. Same stops, same days, same vehicles. Your numbers depend on your stops, windows and fleet; we are happy to run a day of yours.

Build

Everything you click, you can call

The same engines are reachable from your own code. API keys are scoped per capability and can be limited to named agents, flows or tools.

  • Public REST APIAgents, flows, executions, knowledge search and models under /v1.
  • Webhooks inSigned webhook endpoints that start a flow.
  • MCP serverExpose the tools a key allows to any MCP client.
  • Flows as glueCall any HTTP API from a flow step when there is no connector.
# Ask an agent from your own systemcurl -X POST "$API_URL/v1/agents/$AGENT_ID/chat" \  -H "X-API-Key: $API_KEY" \  -H "Content-Type: application/json" \  -d '{     "message": "Which stops are still open on route B?"  }'

Frequently asked questions

Operations software made of engines: Fleet (route planning, tracking, driver app), Agents, Flows, Marketing & Sales, Field service, Comms, Inventory, Invoicing, Reporting and more. They share one contact list, one calendar and one permission model, and you switch on the ones each workspace needs.

No. Engines are switched on per workspace. Many teams start with one, such as route planning or an AI agent, and add others when the work calls for it. Data you already have, like contacts, is shared with whatever you add later.

On nine working days of a Dutch delivery company's real stops, the new solver drove 18,475 km against 19,419 km on the routes the company had saved, and served 1,230 stops against 1,155 with our previous planner, with 70 route crossings instead of 97. On Routific's open benchmark of 121 problems it is within about 1% of Routific on driving distance.

You choose. Agents work with OpenAI, Anthropic, Google, Mistral, xAI, Azure OpenAI and OpenRouter, or with a model you run yourself through Ollama. You can change the model of an agent without rebuilding it.

There are more than 50 connectors, a public REST API with scoped API keys, signed webhooks that start flows, and an MCP server that exposes the tools a key allows to any MCP client. A flow step can call any HTTP API when there is no connector.

On DigitalOcean in EU regions (Frankfurt and Amsterdam). Single-tenant deployment is available on request for teams that need a dedicated environment.

Start with one engine. Add the next when you need it.

Try it free on your own data, or book thirty minutes and we will plan one of your real days with you.

See it running on your own routes

Thirty minutes, screen shared, using a day that looks like yours. Or start free and have a look yourself.

Start free