What ships today and what is next

Available now

• The desktop app for macOS and Windows, and the command line.
• SharePoint Online lists, Microsoft Lists and document-library metadata, from any number of sites.
• Relational, warehouse and lake destinations: SQL Server, Azure SQL, PostgreSQL and its hosted cousins (Azure Database for PostgreSQL, Supabase, AlloyDB, Amazon Aurora, Amazon RDS, Google Cloud SQL, Heroku Postgres, Neon), MySQL, MariaDB, Oracle, SQLite, BigQuery, Databricks, Azure Data Lake (ADLS Gen2), Delta Lake, Apache Iceberg, ClickHouse, DuckDB and Parquet.
• Search and vector destinations: Azure AI Search and Postgres with pgvector, both with embeddings generated on sync from OpenAI, Azure OpenAI or a local model; Elasticsearch, OpenSearch, Qdrant, Weaviate and LanceDB for keyword and filtered search, with no vector.
• Dataverse, for Power Apps that need delegation.
• Full load, then a delta every few minutes, reconcile weekly by default (or on demand with the command-line tool), soft deletes by default (hard deletes with the command-line tool), automatic schema evolution, whole-site selection, per-site least-privilege permissions.
• Desktop app scheduled syncs can keep running with the window closed (macOS launchd agent, Windows Scheduled Task), toggled from Settings.
• An admin MCP server in the SPSync server and desktop app, so an AI assistant can set up connections, destinations and replications and start runs. Secrets go in through a one-time link and never through the conversation, tokens are scoped and expire after 90 days by default, and you can choose 30 or 365 days or never, and every tool call is logged. Set-up is documented for Claude Code, Claude Desktop, Cursor, VS Code and Copilot Studio. See AI and agents.
• A read-only MCP server over the replica, spsync mcp serve, with SharePoint-aware tools (site, list and lookup descriptions).

The destinations page shows the current state of every destination individually.

Next

• SPSync in your Azure, a managed application for your own subscription, is deployed with you on a call while the Marketplace listing is prepared. See SPSync in your Azure.
• Snowflake, Microsoft Fabric (Lakehouse and Warehouse) and Azure Synapse Analytics.
• OAuth sign-in for the admin MCP server, the second stage of that server. Claude.ai and ChatGPT custom connectors need it to reach a hosted SPSync, so neither is supported today. No date is set.

Two-way sync

Writing changes from the destination back to SharePoint, with explicit per-column rules and conflict handling, as an opt-in per replication. Today SPSync is strictly read-only against SharePoint. The write-back design starts with a staging table of proposed changes that an agent or an internal app inserts and a person approves; nothing is scheduled, and telling us which list you would write to is how it moves.

Further out

Amazon Redshift, further out than Snowflake, Fabric or Synapse with no schedule set. Tell us if it is what you need and we will prioritise it.

Under consideration

• On-premises SharePoint 2016, 2019 and Subscription Edition. Under consideration; register interest on the on-prem page.
• Attachment and library file bytes. Metadata is replicated today; file content to blob storage is under consideration.
• Version history replication.
• Column descriptions and a generated schema document per replication, for text-to-SQL tools and agents.
• Embeddings written to Elasticsearch, OpenSearch, Qdrant, Weaviate and LanceDB, as pgvector and Azure AI Search receive today.
• SharePoint item permissions carried into the replica and the search index, so retrieval can be trimmed per user. Today the database role and row-level security set the boundary.