ThinkVoice Public API
Welcome to the ThinkVoice Public API.
The ThinkVoice Public API lets external systems securely query call records in ThinkVoice, download call recordings, manage voice agent calls, and send text channel conversations to ThinkVoice.
How It Works
Using the API comes down to two basic steps:
- Get a Bearer token with your username and password
- Call the relevant endpoint with that token
Available endpoints:
- Call Records — Query analyzed call records
- Call Detail — Retrieve all information about a single call
- Call Recording — Download the audio recording of a call
- Start Voice Agent Call — Start a new call through a voice agent
- Call Logs — List the voice agent call history
- Call Ingestion — Forward a call from your PBX, together with its recording, to ThinkVoice for analysis
- Text Channels — Forward a conversation from a text channel such as live chat, together with its messages, to ThinkVoice for analysis
ThinkVoice can also notify your system: when a call analysis completes, it automatically sends a notification to the URL you configure. See the Webhook section for details.
The tenant is not selected via a header. Only the data of the tenant the user belongs to is visible.
Key Facts
| Property | Detail |
|---|---|
| Base URL | https://public-api.thinkvoice.ai |
| Auth Method | Bearer Token |
| Token Lifetime | Access token 2 hours (7200 seconds), refresh token 7 days |
| Rate Limit | General: 40 requests / 60 seconds · Token: 5 requests / 60 seconds · Call Ingestion and Text Channels: 600 requests / 60 seconds |
| Data Source | Elasticsearch (per-tenant indices) |
Security
- Only users granted Public API access can obtain a token
- An optional IP restriction can be applied
- Sensitive data in call titles and summaries is returned as
[MASKED]— phone numbers, email addresses and Turkish national ID numbers (TCKN) are masked - The customer phone number is returned only in the
customerPhoneNumberfield; it never appears as plain text in text fields
Before You Start the Integration
Before starting the integration, you need to share the following information with our team:
| # | Information | Description |
|---|---|---|
| 1 | Environment | production |
| 2 | User credentials | Email and password of the user who will call the API (Public API access is granted to the user by our team) |
| 3 | Outbound IP addresses | If IP restriction will be enabled, as static IPs or in CIDR format |
| 4 | Call type filters | Full names of any types to be used other than the defaults |
| 5 | Date/time convention | Whether UTC or local time will be used |
| 6 | Expected traffic volume | Rate limits can be reviewed for high volumes |
| 7 | Call ingestion (PBX integration) | If you will send calls from your PBX, you will receive X-Api-Key and X-Api-Secret values |
Practical Tips
- Run your first tests with cURL
- Get a token first, then test the endpoints one by one with the same token
- The token is valid for 2 hours; renew it with
refresh_tokeninstead of logging in on every request (the token endpoint is limited to 5 requests per minute) - If you need deep pagination, keep the
skip + take ≤ 10,000limit in mind (for/public-calls/get-calls) - You can get the latest state of a call by querying the
callIdreceived via webhook with/public-calls/call - You can track calls started with
voice-agent/start-callthroughvoice-agent/call-log - Do not treat feedback summary texts as raw data — sensitive fields are masked
Last updated: