Turn recorded sales calls into clear insights your team can act on. The flow captures audio, turns it into text, scores the conversation, and saves a structured summary to your database. Sales leaders and reps can see interest, intent, objections, and upsell signals without manual note taking.

Under the hood, an audio link is sent to a transcription service using a manual test path. A webhook receives the callback when the transcript is ready. The flow checks the status, fetches the full text, and sends it to an AI model with a specific prompt and JSON schema. The result is a clean JSON object with fields like interest score and next steps. Finally, the data is written to a Supabase table so you can query it or show it in dashboards.

Setup needs API keys for the transcription tool, the AI model, and your database project. Replace the audio link, set your webhook URL, and adjust the prompt and schema to match your sales criteria. Expect a big drop in review time and faster coaching loops. Ideal for outbound call reviews, discovery calls, and tracking upsell opportunities at scale.

• Manual test path sets a direct audio URL for quick runs and debugging.
• Starts transcription via AssemblyAI with speaker labels and expected speakers.
• Receives status updates through a webhook and checks completion before moving on.
• Fetches the final transcript text using the transcript ID from the webhook payload.
• Builds a clear analysis prompt and JSON schema with a Set node for consistent output.
• Calls OpenAI chat completions to return structured insights like scores and next steps.
• Saves both the raw input and parsed output into a Supabase table for reporting.

• Reduce manual call review from 60 minutes to 5 minutes
• Automate 90 percent of note taking and scoring
• Improve data accuracy by 95 percent with structured JSON
• Handle 10 times more calls with the same team
• Connect transcription, AI analysis, and database in one flow

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with AssemblyAI, OpenAI and Supabase. See the Tools Required section above for links to create accounts with these services.
3. In n8n, open the AssemblyAI HTTP Request nodes. In the credential dropdown, click Create new credential. Choose HTTP Header Auth and add an Authorization header with your AssemblyAI API key from the AssemblyAI dashboard. Name the credential clearly, for example AssemblyAI Prod.
4. Open the OpenAI node. In the credential dropdown, click Create new credential and paste your OpenAI API key from the OpenAI API page. Name it OpenAI Prod. Keep the model as configured unless you have cost or policy reasons to change it.
5. Open the Supabase node. In the credential dropdown, click Create new credential and enter your Supabase project URL and API key. Use a key that allows insert on the demo_calls table. Name it Supabase Prod.
6. Open the Set link to audio node and replace the sample URL with a direct, publicly reachable link to your audio file. If the link needs headers or login, it will fail, so use a direct file URL.
7. Open the AssemblyAI Transcribe node body and set the webhook_url to the Production URL of your n8n Webhook node. Copy the Production URL from the Webhook node panel. Save the node.
8. Open the Set vars node and adjust the prompt text and JSON schema details to match the fields you want, such as interest score, objections, and next steps. Keep fields aligned with your Supabase table structure.
9. Click Execute on the Webhook node so it is ready to receive the AssemblyAI callback. Then run the manual trigger path to start a new transcription request.
10. Wait for AssemblyAI to finish. The If node allows the flow to continue only when the status is completed. Watch the Execution list for the webhook call, then confirm the OpenAI step runs and the Supabase record is created.
11. Open your Supabase project and check the demo_calls table. Verify the input text, audio URL, transcript ID, and the parsed analysis JSON are saved as expected.
12. Troubleshoot common issues: 401 errors mean a bad API key, missing transcript usually means the webhook URL is wrong, and empty text often means the audio link is not publicly accessible. If OpenAI fails, simplify the prompt or remove fields from the JSON schema and try again.