Skip to content
Docs

Set up Callario, step by step.

From an empty workspace to an AI agent answering on your website, with your team, routing and workflows behind it. During the alpha every workspace is set up with you — so read this as the map of that work, and of where each setting lives.

Jump to

01Getting started

Seven steps
to a live AI agent.

In the order teams usually do them. The console keeps its own checklist on Home, which ticks each one off as it’s really done.

  1. Step 1

    Your workspace and team

    During the alpha your workspace is created with you at onboarding. Sign in with your email and password, turn on two-factor authentication, and invite teammates with a role each — an invite counts once it’s accepted.

    • Settings → Team
    • Settings → Security
    Security & trust

  2. Step 2

    Add your help-centre articles

    Write articles in Knowledge, or import HTML, Markdown or PDF files. The writing assistant can improve, shorten, simplify or draft from a description, in English or Arabic. Published articles are what your AI agent answers from, and what customers search in your help-centre portal.

    • Knowledge → New article
    • Knowledge → From file
    • Settings → Portal
    Knowledge & Help Centre

  3. Step 3

    Create and test your AI agent

    Create an agent from a starter template, describe what it should handle, and choose the knowledge it answers from. Talk to it in the test chat before your customers do; an evaluation gate can hold it back from publishing until it passes.

    • AI Agents → New agent
    AI AgentGuardrails

  4. Step 4

    Put it on your website

    Publish the agent, then copy one script tag from its Deploy tab into your site — or share the hosted chat link, which needs no code at all. Both speak English and Arabic, right to left.

    index.htmlhtml
    <!-- paste before </body> on every page that should show the widget --><script src="https://api.callario.com/widget.js"  data-agent-key="pk_live_…" defer></script>
    • Your agent → Deploy
    AI Agent

  5. Step 5

    Decide when a person takes over — and who

    Set the agent’s handoff: when the customer asks for a human, gets frustrated or goes in circles, when the agent is unsure, after a number of turns, or outside office hours. Then build queues with a strategy — least busy, round robin, longest idle or manual — plus skills and capacity, and add SLA policies with business hours.

    • Your agent → handoff
    • Inbox → Queues
    • Settings → SLA
    Routing & SLAsInbox & Ticketing

  6. Step 6

    Automate the rest with workflows

    Describe a workflow in plain language and AI drafts it, or start from a template such as “Escalate a missed SLA” or “AI drafts, a teammate approves”. Test-run it with no side effects, publish it, and follow every run step by step.

    • Workflows → New
    • Workflows → Templates
    Workflows & Rules

  7. Step 7

    Connect your own systems (optional)

    Create a scoped API key, subscribe an HTTPS endpoint to signed webhooks, or register a context endpoint so the agent can look up a verified customer’s order or account on your side.

    • Settings → Integrations
    API referenceIntegrations & API

02Reference

For developers
and the curious.

API referenceBase URL, keys and scopes, the four endpoints, errors and rate limits.Webhooks & eventsSignature verification, retries and replay, and the event catalogue.GlossaryThe terms used across the product and this site, defined plainly.Security & trustHow data is isolated and protected — and what isn’t in place yet.ChangelogWhat shipped, month by month.Talk to usStuck on a step? During the alpha, setup is done with you.

03Module by module

What each part does,
and what it doesn’t yet.

Each product page covers one module in depth — its screens, what the AI does there, and a plain FAQ.

Design-partner programme · now onboarding

Rather set it up together?

Design partners get every step above done with them, on their own articles and real questions.

Get early accessAbout design partners