> ## Documentation Index
> Fetch the complete documentation index at: https://docs.teamfollowup.ai/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> Use the public API base URL: https://api.teamfollowup.ai/api.
> Authenticate public API requests with Authorization: Bearer YOUR_API_KEY.
> Use product-owned terms: agents, campaigns, projects (GoHighLevel sub-accounts), contacts, calls, outcomes, skills, cadences, phone numbers.
> Scope requests by locationId rather than projectName.
> Read the guide linked from each API reference group before recommending an endpoint.

# Your first script

> Co-author asks a few short questions about your business and your offer, then writes a first draft for you to review.

Co-author writes your first script from a short conversation. It asks only what
it does not already know, one question at a time, then shows you a draft with
the assumptions it made.

## What a good first script needs

* **What your business does.** If your knowledge base already says this,
  Co-author will not ask.
* **Why you are contacting these people.** Where they came from, or what they
  did: answered an ad, filled in a form, booked an appointment.
* **What they are being offered, in plain terms.** "A free 20 minute roof
  assessment and a written quote within a day" gives the agent something to say.
  "A consultation" does not, so expect one follow up question.
* **Who is worth booking, when that matters.** If everyone who is interested
  should be booked, say so or skip it. If a visit costs you real time, tell
  Co-author who to screen out.

Inbound campaigns are simpler: tell Co-author what people usually call about.
Confirmation campaigns need to know whether appointments are online, in person,
or both.

## In a hurry?

Say "just build it" at any point. Co-author builds with what it has and lists
what it assumed, so you can correct the draft instead of answering questions.

## Examples and useful details

<AccordionGroup>
  <Accordion title="Describe what your leads get from the call">
    A useful answer explains the benefit in everyday language. Include any
    price, commitment or timing you know. Co-author uses your own description
    to make the script specific to your business.
  </Accordion>

  <Accordion title="Confirm online or in person appointments">
    Online appointments let the agent point people to their joining
    instructions. In person appointments let it help them prepare for the
    visit. Co-author uses the details you give and never invents an address or
    a joining link.
  </Accordion>

  <Accordion title="Identify the lead source">
    Examples include a Facebook ad, a website form, a referral, or an existing
    customer list. The source is how the agent explains why it is calling.
  </Accordion>
</AccordionGroup>

## Choose the conversation and its capabilities

You decide the purpose and wording of the script. [In-call capabilities and
tools](/skills/overview) provide actions such as booking an appointment,
transferring a call, or adding a tag. An agent can also have no installed
capabilities.

Asking for a booking in the script does not install booking. Configure the
capability and its calendar when you want the agent to book during a call.
Your campaign's trigger separately controls how the conversation starts.

## Already have a script?

Paste it into the editor and tell Co-author what you want to change. You can
also say “I already have a script” or “skip the setup questions” to take this
path. Once the script is in place, Co-author can help you refine it.

## Finding the setup details

You can answer in your own words. If you want help, ask Co-author to check your
connected HighLevel account for clues about your lead sources. It asks for your
agreement before scanning and brings any suggestion back for you to confirm.

On a sub-account without a CRM there is no HighLevel account to check, so
Co-author asks you directly. Booking still works there with a TFU AI, Cal.com
or Calendly calendar.

Anything you have already said counts. Co-author should never ask you the same
thing twice.

## Review your first draft

The draft contains **Context**, [Steps to Follow](/agents/writing-steps), and a
[Conversational FAQ](/agents/writing-faq). Review the script and its assumptions,
then tell Co-author what to adjust.

You can change a step, add an FAQ entry, or update [how your agent
sounds](/agents/voice). To generate a fresh draft later, explicitly ask to
rebuild the script and review the proposed replacement.

## In the API

* [Create agent](/api-reference/agents/create-agent) is where a new agent starts
* [Get agent](/api-reference/agents/get-agent) with `view=script` reads the draft back
* [Update agent](/api-reference/agents/update-agent) saves the edited script as `prompt`; see [The prompt](/agents/prompt)
