Summary
Highlights
This video introduces a more efficient way to use AI by building automated workflows instead of manual copy-pasting. It focuses on connecting OpenAI with n8n to create custom AI prompts and automate tasks like transcribing audio files.
To get started, you need an OpenAI developer platform account (not a ChatGPT profile). Create an account, add payment info, and acquire credits. Then, authorize n8n to access your OpenAI account by creating a credential using an API key generated from your OpenAI settings.
The tutorial demonstrates building an automation using AirTable as a trigger. The workflow runs when a new record is added to an 'audio transcripts' table, sending the attached audio file to OpenAI for transcription and summary. Ensure 'download attachments' is enabled in the AirTable trigger settings.
Add an OpenAI step to the workflow and select 'Audio actions' > 'transcribe a recording'. Choose your OpenAI credential and provide the input data field name containing the audio file. Executing this step will send the audio to OpenAI for transcription, costing a minimal amount per minute.
To summarize the transcription, add another OpenAI step and choose 'message a model'. Select your OpenAI account, pick a model (e.g., GPT-4o), and compose a prompt like "Summarize this transcript in one to two sentences." Drag the text variable from the previous transcription step into the prompt.
Finally, add an AirTable step to update the original record with the AI-generated transcription and summary. Select 'update record' as the action, choose the correct base and table, and map the AI responses to the 'transcript' and 'summary' fields. Test the step to ensure the data is updated in AirTable.
Enable the automation in n8n. The workflow will now automatically check AirTable for new records, transcribe and summarize audio files, and update AirTable accordingly, allowing AI to run seamlessly in the background without manual intervention.