Meeting Summary
The Meeting Summary template automatically transcribes an audio recording and generates a structured summary. It uses AssemblyAI to convert speech to text and OpenAI to produce a clean, readable summary of the meeting.
How the flow works
_10AssemblyAI (Start Job) → AssemblyAI (Poll) → Parser → Prompt → Language Model → Chat Output_10 ↓_10 Memory History → Prompt → Language Model → Chat Output
The AssemblyAI Start Transcript component submits the audio file for transcription. The AssemblyAI Poll component waits for the transcription job to complete and retrieves the result. A Parser extracts the raw transcript text, which is passed through two LLM stages: one to produce an initial summary, and one to refine it into a final polished output. Memory stores intermediate results between stages.
Prerequisites
- An AssemblyAI API key — used to transcribe the audio
- An OpenAI API key — used to summarize the transcript
- An audio recording (most common formats are supported: MP3, MP4, WAV, M4A, FLAC, OGG, WebM)
Quickstart
-
In Orbis Studio, open the Templates library and select Meeting Summary.
-
In the first AssemblyAI component (Start Transcript), add your AssemblyAI API key and upload your audio file.
-
In the second AssemblyAI component (Poll), add your AssemblyAI API key.
-
In both Language Model components, add your OpenAI API key (or select a different provider and model).
-
Open the Playground and click Run. The flow transcribes the audio and returns a summary.
Key components
| Component | Role |
|---|---|
| AssemblyAI Start Transcript | Submits the audio file to AssemblyAI and starts a transcription job |
| AssemblyAI Poll | Waits for the job to complete and retrieves the transcript text |
| Parser | Extracts the transcript text from the AssemblyAI response |
| Prompt (×2) | Formats the transcript and structures the summarization task for the LLM |
| Language Model (×2) | Generates and refines the meeting summary |
| Memory | Stores intermediate outputs across processing stages |
| Chat Output (×3) | Displays the transcript and summary at each stage |
Configuration tips
- Audio format: AssemblyAI supports MP3, MP4, WAV, M4A, FLAC, OGG, and WebM. For best results, use clear audio with minimal background noise.
- Summary style: Edit the Prompt templates to customize the output — for example, add "include action items and decisions" or "format as bullet points."
- Long recordings: AssemblyAI handles long files, but processing time scales with audio length. For very long recordings, consider splitting the audio beforehand.
- Language: AssemblyAI supports multiple languages. Check AssemblyAI's documentation for the full list and how to set the language parameter.
Next steps
- Store summaries to a database by adding a Write File or external integration component at the end of the flow.
- Add a Knowledge Ingestion step to index meeting summaries in a Knowledge Base for later search.