# Workspace Configuration

# Workspace Configuration

Workspace configuration covers the settings that control how your environment behaves: languages and NLP defaults, tokens and integrations, search/index settings, and operational parameters.

This page focuses on practical defaults and the configuration surfaces you will use most often.

# Core configuration areas

• Workspace identity
  • name and branding used in the UI
  • environment naming (dev/staging/prod)
• Languages
  • select the languages present in your data
  • ensure NLP pipelines are enabled for the languages you care about
• Authentication and tokens
  • admin accounts and user authentication
  • API tokens for connectors/integrations
  • endpoint tokens for external access to custom endpoints
• Data and schema
  • create/update node and edge schemas
  • re-index/re-parse operations when schemas or pipelines change
• Search
  • text indexes over chosen fields
  • vector/embedding indexes where semantic retrieval is needed
  • facets and filters configuration

# Recommended initial setup checklist

• Set workspace name and validate the base URL your users will use.
• Configure languages for your data (don’t enable languages you don’t need).
• Create an API token for ingestion connectors and store it in your secret manager.
• Create an endpoint token if you will call custom endpoints from outside the UI.
• Decide indexing boundaries early:
  • which node types are searchable?
  • which fields are indexed as text vs embeddings?
  • which fields become filter facets?

# Operational knobs you will use frequently

• Re-index / rebuild search when changing search configuration or indexed fields.
• Re-parse when changing NLP pipelines or entity capture/linking configuration.
• Export/import configurations (recommended for promoting changes from dev → prod).

# Common pitfalls

• Over-indexing: indexing every field increases storage and can degrade relevance; start with the user-facing fields.
• Mixed identifiers: if keys aren’t stable, connectors will create duplicates; choose stable IDs or deterministic hashes.
• Unclear access control: define permission boundaries early for production environments.

# Next steps

• Integrate data: Data Integration → Connectors
• Configure search: Search → Text Search
• Configure NLP and embeddings: NLP → Overview