Nixflex - AI Voice Platform for Building AI Phone Agents

This guide builds an outbound agent that calls your leads, asks a few qualifying questions, and returns the answers as structured data — ready to push into your CRM or review in the dashboard.

What you’ll build

An outbound agent that:

Calls each lead with their details already personalized into the conversation
Asks your qualifying questions naturally
Captures the answers as structured fields after every call

Before you start

An agent created and a number attached
Your API key
A list of leads with the details you want to personalize (name, company, anything relevant)

Steps

Personalize each call with variables

Pass a lead’s details as variables when you start the call. The agent can use them in its prompt so every call feels tailored.

curl -X POST https://api.nixflex.com/v1/calls \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agent_id": "agent_xxx",
    "to": "+15551234567",
    "dynamic_variables": {
      "lead_name": "Jordan",
      "company": "Acme Ltd"
    }
  }'

In the agent prompt, reference them with double curly braces:

Agent prompt (excerpt)

You are calling {{lead_name}} from {{company}} on behalf of Nixflex.
Introduce yourself warmly, confirm you are speaking to the right person,
and ask if now is a good moment before you continue.

Ask your qualifying questions in the prompt

Tell the agent what to find out. Keep it conversational — the agent works the questions into a natural call.

Agent prompt (excerpt)

Your goal is to find out three things:
- Whether they are the decision maker
- Their rough budget
- Whether they want a follow-up call

Ask naturally, one question at a time. Do not read them as a list.
Thank them and end the call politely once you have what you need.

Set the agent’s response_length to short so replies stay crisp on a sales call. See Agents for the behaviour settings.

Capture the answers as structured data

Define an analysis schema so every call returns the qualifying answers as clean fields — not just a transcript. The engine fills these in after the call ends.

{
  "analysis_schema": {
    "decision_maker": { "type": "boolean", "description": "Is the person the decision maker?" },
    "budget": { "type": "string", "description": "The budget range they mentioned" },
    "wants_followup": { "type": "boolean", "description": "Do they want a follow-up call?" }
  }
}

After each call, read the results from the call record — the structured fields sit alongside the summary and sentiment.

Scale to your whole list

Repeat the call request for each lead, passing that lead’s variables each time. Every call personalizes itself and returns the same structured fields, so your results line up into one clean table.

What’s next

Read Outbound calls for the full call API and options.
See Variables for everything you can personalize.
Use Post-call analysis to shape exactly which fields you capture.

Outbound calls

Starting calls and passing lead details through the API.

Post-call analysis

Turning each call into structured, reviewable data.

Get started

Guides

Best practices

Core concepts

In-call actions

Features

SMS

Reference

Outbound lead qualification

What you’ll build

Before you start

Steps

What’s next

Outbound calls

Post-call analysis

​What you’ll build

​Before you start

​Steps

​What’s next

Outbound calls

Post-call analysis

What you’ll build

Before you start

Steps

What’s next