Skip to main content

Getting Started

Getting started with Agentic takes one environment and one published agent. Nothing in Settings is needed first: agents are built, tested and published without Twilio, and a phone number or human escalation can be added later, or never. This page covers access, turning the product on, the first environment, and the Getting started checklist that tracks the rest. The Quickstart then walks through the first agent.

Before You Start​

  • You need the Admin or Organisation Admin role in the organisation. No other role can open Agentic.
  • To give a Frontline agent a phone number later, the organisation needs a Twilio account. Twilio is the telephony provider that supplies phone numbers and carries the calls. An Organisation Admin connects it in Settings.

Who Does What​

RolePart in getting started
AdminTurns Agentic on, creates environments, and builds, tests and publishes agents
Organisation AdminEverything an Admin does, plus Settings: connecting Twilio, setting up human escalation, and managing every connected app in the organisation

An Admin who reaches a step that needs Settings sees a Copy link for admin button instead of a link. It copies the address of the right settings page, ready to send to an Organisation Admin. See Roles and Permissions for the full list.

Step 1: Turn On Agentic​

Agentic is turned on for an organisation by creating its first environment. There is no separate request to make.

  1. Log in to the NeonNow Admin app as an Admin or Organisation Admin.
  2. Select Agentic in the sidebar.
  3. On the product landing page, select Get started.

The landing page only appears while the organisation has no environments. Once one exists, selecting Agentic in the sidebar opens the Agents page of an environment instead.

Step 2: Create the First Environment​

Get started opens Set up your first environment. An environment keeps a set of agents and their settings together, so things can be tried in a sandbox before they reach customers.

  1. Enter a Name, for example "Clinic pilot".
  2. Choose a Type. Sandbox is selected by default, which suits a first environment for trying things out.
  3. Optionally, select Add a description and say what the environment is for.
  4. Select Create Environment.

The new environment opens on its Agents page, ready for the first agent. See Creating an Environment for what each type is for, and how to add more environments later.

Step 3: Build, Test and Publish an Agent​

The first agent is built from a plain description. IQ drafts it, asks a few questions to fill the gaps, and hands over a draft ready to test. The Quickstart walks through it end to end.

The Getting Started Checklist​

While an environment still has setup left to do, a Getting started control sits at the bottom of the Agentic sub navigation, with a progress bar and a count of the steps done so far. Selecting it opens What's Left in This Environment, a short checklist with a button beside each step that is waiting on you.

A status line above the list summarises the Twilio connection, for example "Twilio not connected · agents work without it".

StepRequired?What finishes it
Build an AgentYesCreating the first agent in the environment, with New Agent
Test ItYesTesting an agent in the playground, or publishing one
Publish ItYesPublishing an agent. Once published, your own applications can reach it by its agent ID, and a phone number is optional
Give It a Phone NumberOptionalConnecting Twilio and assigning a number to an agent
Add Human EscalationOptionalTurning on human escalation, connecting Twilio, and pointing Twilio at the environment. See The Human Escalation Step

The checklist is worked out from what the environment already has, so it updates on its own as each step is done. Once nothing is outstanding the control disappears. An environment that never needs a phone number or human escalation is finished as soon as an agent is published.

The Human Escalation Step​

The last step changes its title as human escalation is set up. Each title is a separate state of the same step:

TitleWhat it meansButton
Add Human EscalationHuman escalation is off. Turning it on lets agents hand conversations to your team, and needs a Twilio API key and TaskRouter, Twilio's tool for routing work to peopleSet Up, which opens Human Escalation
Finish Human EscalationHuman escalation is on, but Twilio is not connected or its saved details do not work. Human escalation runs on the organisation's Twilio account, so that comes firstNone. The step reads After Twilio until the Twilio connection works
Point Twilio Back at UsTwilio is connected, but the addresses Twilio uses to send escalated calls and chats to this environment are not set up yet, or some now point somewhere else. When they point elsewhere, the step shows an amber warning, because that traffic is not reaching the environmentReview, which opens the webhook addresses in Twilio Resources
Human Escalation ReadyEverything is in place. The step sums up the setup, for example "3 people are in the workspace · all 4 webhooks in sync"None

Add Human Escalation stays marked Optional and does not hold the checklist open. Once human escalation is turned on, the step counts as outstanding until it reaches Human Escalation Ready. For an Admin, Set Up and Review appear as Copy link for admin, because these pages are in Settings.

When the Checklist Comes Back​

The control returns if something that was working breaks. The step that needs attention shows an amber warning:

  • Twilio Rejected Your Saved Credentials: the Twilio details are saved but Twilio refuses them. Select Fix Connection to correct them in Twilio settings.
  • Point Twilio Back at Us: some of the addresses Twilio uses to reach the environment point somewhere else, so escalated calls and chats are not arriving. Select Review to check them. See The Human Escalation Step.

An optional step that has been started also stays on the list until it is finished, for example Finish Human Escalation after human escalation is turned on.

Agents work without Twilio

Twilio is only needed for a phone number on a Frontline agent, or for human escalation. Building, testing and publishing work without it, and a published agent can already be reached by its agent ID. See Getting Started in the developer docs.

What Happens Next​