Documentation
What Is Starmesh?
Starmesh is a unified analytics platform that transforms raw unstructured data — like call transcripts, meeting recordings, chats, social media posts, and documents — into structured, analyzable insights.
We support seamless ingestion from multiple platforms (Zoom, Teams, Gong, Fireflies, Reddit, Google Drive, etc.) and apply advanced AI to extract topics, sentiment, entities, and themes — all without requiring manual effort or custom code.
Key Capabilities
- Multi-source ingestion: Connect to audio/video platforms, chat systems, forums, document storage, and more.
- AI-powered transforms:
- Topic modeling (guided or unsupervised)
- Sentiment analysis (overall, per speaker, per topic)
- Entity recognition and classification
- Embedding generation and semantic search
- Pipeline builder: Define reusable data flows from source to transform to storage.
- Export to warehouse: Write output to BigQuery, GCS, or download as CSV/JSON.
Connecting a Data Source
Starmesh supports multiple types of sources. Here's a high-level process to connect one:
- Go to your dashboard → Sources.
- Select the source type (e.g., Zoom, Reddit, Teams, Google Drive).
- Follow the prompts to authenticate (via OAuth, service account, or API token).
- Choose the subset of data you want to analyze (e.g., folders, users, topics, recordings).
- Starmesh will begin ingesting and preprocessing the content.
Supported source types include:
| Type | Examples |
|---|---|
| Audio/Video platforms | Zoom, Microsoft Teams, Gong, Fireflies |
| Social & forums | Reddit, Twitter/X, LinkedIn |
| Messaging & support | Intercom, Slack, Salesforce cases |
| File & doc storage | Google Drive, Dropbox, SharePoint |
| Email (upcoming) | Gmail, Outlook |
Note: Not all integrations are available yet. Some source types are coming soon!
Built-in Transforms
| Transform | Description |
|---|---|
| Topic Modeling | Extract structured topics from conversations or text (LLM or traditional models) |
| Sentiment Analysis | Detect sentiment per document, per speaker, or by topic |
| Entity Recognition | Identify companies, products, features, competitors, people, and intent |
| Embeddings & Search | Generate dense vector representations for semantic similarity or retrieval |
| Custom Prompt Workflows | Define prompt-based transformations with output schema enforcement |
Destinations
Once analyzed, your data can be exported to:
- Google BigQuery
- Google Cloud Storage
- Download
- Upcoming: Snowflake and PostgreSQL
Testing and Observability
- Preview mode for pipeline testing
- Logs and data flow diagrams
- Versioned runs to compare model configurations
Security and Access Control
- OAuth 2.0 and service account authentication
- Encryption at rest and in transit
- Fine-grained access control
- No indefinite storage of raw content unless explicitly configured
Disconnecting a Source or Deleting Data
To disconnect a data source:
- Go to Sources tab.
- Click the settings icon for the source.
- Click Revoke Access or Disconnect.
To delete ingested data, go to Settings → Data Management and request deletion, or email support@starmesh.ai.
Removing Starmesh from your Zoom Apps
- Log in to your Zoom account and navigate to the Zoom App Marketplace.
- Click Manage >> Added Apps or search for the "Starmesh" app.
- Select the "Starmesh" app.
- Click Remove.
Need Help?
Email: support@starmesh.ai
Website: https://starmesh.ai