Docs
Get started

Wazzap meets OpenAI

Drop your OpenAI API key into Wazzap and unlock AI everywhere: smart auto-replies, voice-note transcription, message summaries, and the GPTalk Pro assistant. One key wires up every sub-account you manage.

Overview

OpenAI is the brain behind several Wazzap features. You bring your own API key so the cost, the model choice, and the usage limits stay under your control. Wazzap never proxies billing, we just call OpenAI as if it were you, with the credentials you provide.

★

You only need to connect OpenAI once at the sub-account level. Every feature that uses GPT, Whisper, or TTS will pick up the same key automatically.

Requirements

  • An OpenAI account with billing enabled. Free tier won't work for production sends.
  • An API key from platform.openai.com/api-keys.
  • A funded balance. Even $10 USD lasts a while at GPT-4o-mini prices.

Connect Wazzap to OpenAI

1. Generate an API key

  1. Log in to platform.openai.com.
  2. Open API Keys in the sidebar.
  3. Click Create new secret key. Name it something like "Wazzap, production".
  4. Copy the key. It starts with sk-. You'll only see it once.

2. Paste it into Wazzap

  1. Open your Wazzap Sub-Account Panel.
  2. Go to Settings, General.
  3. Paste the key in the OpenAI API key field.
  4. Click Save. Wazzap validates the key against OpenAI and shows a green check.

3. Pick a default model

Choose which model Wazzap uses when a feature doesn't specify one. gpt-4o-mini is the recommended default, fast, cheap, and good enough for chat replies and summaries. Switch to gpt-4o or newer if you need higher quality.

What you can do with it

  • GPTalk Pro assistant. A drop-in AI agent that reads WhatsApp threads and replies on your behalf. See the GPTalk Pro guide.
  • Auto-reply with prompts. Define a system prompt per sub-account and let GPT handle the first response.
  • Voice-note transcription. Wazzap converts incoming voice notes to text with Whisper and pushes the transcript into your CRM.
  • Text-to-voice. Use OpenAI's TTS to ship voice notes generated from text, via the voice command.
  • Conversation summaries. Pin a long thread and get a one-paragraph TL;DR on demand.

Voice notes and transcription

Two voice-related capabilities run on OpenAI by default:

  • Inbound transcription (Whisper). Every voice note your number receives is transcribed and stored alongside the audio in the conversation. Toggle it on under Settings, Send transcription.
  • Outbound voice (TTS). Send a command like #voiceAI|OAI|alloy|Hi there, this is Diego and OpenAI generates a voice note in one of its preset voices. For ElevenLabs voices, see the ElevenLabs integration.

Notes and limits

  • Keep your API key active and funded. An expired or zero-balance key breaks every AI feature on that sub-account.
  • OpenAI enforces per-minute rate limits. If you saturate them with bulk replies, calls will queue and eventually fail. Upgrade your OpenAI usage tier if needed.
  • Wazzap caches model lists locally and refreshes them daily. New models show up within 24 hours of OpenAI launching them.
!

Treat your key like a password. If anyone with access to your team accidentally commits it to a repo, rotate it from platform.openai.com and update it in Wazzap.

Troubleshooting

"Invalid API key" on save

You pasted the key with leading whitespace or you copied the OpenAI organization ID by mistake. Generate a fresh key and paste it without extra characters.

Replies come back empty or as plain "…"

Either your OpenAI balance hit zero, or you hit the per-minute rate limit on your usage tier. Top up your billing and try again. The Wazzap log under Settings, Activity shows the exact OpenAI error code.

Transcripts don't appear

Confirm Send transcription is enabled on the sub-account, and that the Whisper feature is available in your OpenAI account region. Voice notes longer than 25 minutes are split internally before being sent to Whisper.

Was this page helpful?