Docs

HEYRIK DOCUMENTATION

Start building with HeyRik.

Turn a business task into a useful conversation. Build your first voice agent in the dashboard, or connect HeyRik to your own application.

01

From first draft to first call

A clear sequence, with a useful result at every step.

  1. Define the job

    Choose one outcome and write a clear brief.

    Plan your agent
  2. Build the agent

    Create a draft. Add its instructions, voice and knowledge.

    Create a draft
  3. Test the conversation

    Try real questions, difficult cases and human handoffs.

    Test and launch
  4. Learn from every call

    Review the results and improve one thing at a time.

    Review call results
02

The building blocks

Each part has one clear responsibility.

03

What you can build

The same platform primitives support different industries and outcomes.

AI receptionist

Answer every inbound call, explain opening hours, collect the caller’s intent, book appointments, and route urgent requests.

Lead qualification agent

Call new leads, understand requirements, qualify interest, capture structured fields, and schedule the right follow-up.

Customer support agent

Identify the customer, answer from approved policies, look up information through a custom API, and create a ticket when needed.

Campaign calling agent

Call a list of leads for reminders, surveys, renewals, or outreach, then export outcomes for your workflow.

04

Create your first agent

This small example creates a support agent. The request defines its identity, opening line, and basic behavior.

Request
curl -X POST https://server.heyrik.com/api/v1/agents \
  -H "Authorization: Bearer $HEYRIK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Support Agent",
    "language": "en-IN",
    "greeting": "Hi, thanks for calling Acme. How can I help?",
    "config": {
      "personality": "Friendly",
      "prompt_sections": [
        {
          "order": 0,
          "title": "Identity & Purpose",
          "content": "You are Acme’s support agent. Resolve the caller’s question using approved company information."
        },
        {
          "order": 1,
          "title": "Guardrails",
          "content": "Never invent policy details. If the answer is unavailable, offer a human follow-up."
        }
      ]
    }
  }'
Example output
{
  "data": {
    "id": "agc_9f2a1c30",
    "name": "Support Agent",
    "language": "en-IN",
    "greeting": "Hi, thanks for calling Acme. How can I help?",
    "config": {
      "personality": "Friendly",
      "prompt_sections": [
        { "order": 0, "title": "Identity & Purpose", "enabled": true },
        { "order": 1, "title": "Guardrails", "enabled": true }
      ]
    },
    "created_at": "2026-07-24T06:11:44.201Z"
  }
}
05

What the call output looks like

After a call finishes, your application can read the outcome and use it in the next workflow.

Completed support call
resolved
Caller+91 98765 43210
Duration3m 42s
OutcomeRefund policy explained
Extracted fieldorder_id: ORD-2048
Next actionNo follow-up required
GET /calls/{id} response
{
  "data": {
    "id": "call_1dd990f7",
    "status": "completed",
    "outcome": "resolved",
    "summary": "Caller asked about the refund window. The agent explained the approved 14-day policy.",
    "extracted_variables": { "order_id": "ORD-2048" },
    "transcript_url": "https://…",
    "recording_url": "https://…"
  }
}
06

Choose your next step

Continue with the guide that matches what you need to build.