SPSync vs Airbyte

Choose SPSync when SharePoint lists are the whole job and you want lookups as joinable keys; choose Airbyte when SharePoint is one source among many on a platform you already run.

Documentation reviewed 23 October 2026

Where Airbyte fits

• Open source and self-hostable
• very wide connector catalogue
• active community
• Airbyte Cloud managed option
• a published RAG tutorial that loads SharePoint files into a vector database, with chunking and embedding done in the destination connector

Where it breaks

• the SharePoint Lists Enterprise connector is no longer sold, and the Marketplace Microsoft Lists connector syncs by full refresh only
• the Lists Enterprise connector returns lookups as display text and multi-value lookups as semicolon-separated strings, not keys or child tables
• the Lists Enterprise connector does not sync deleted items, and new columns need a schema refresh
• self-hosting means operating the Airbyte platform itself

Feature by feature

If SharePoint is one source among fifty, Airbyte is a sensible place to put it. Teams that already run Airbyte, self-hosted or on Airbyte Cloud, to move Salesforce, Postgres and a dozen SaaS APIs into a warehouse keep one scheduler and one monitoring surface. The open-source core is free, the connector catalogue is broad and the community is active. When the job is “get everything into the warehouse”, that consolidation is worth a lot.

What to check before you rely on the connector

SharePoint lists reach Airbyte by two routes, and both deserve a look. The SharePoint Lists Enterprise connector is the detailed one, and Airbyte’s documentation now says it is no longer sold, only supported for earlier buyers (checked 2026-10-23). Its documentation says lookups return the display value as text, multi-value lookups arrive as semicolon-separated strings, deleted items are not synced and new columns need a schema refresh. The Microsoft Lists connector in Airbyte’s Marketplace is the one you can add today, and every stream it offers is a full refresh. Self-hosting also means operating the Airbyte platform itself.

A schema built from SharePoint’s column types

SPSync replicates each list into a normalised set of tables and keeps them current every few minutes. Lookups become key columns you can join on, multi-value and multi-lookup columns become child tables, and people resolve into a shared sp_users table, so a display name is stored once. Deletes are tracked (soft by default, hard delete with the command-line tool), and a column added in SharePoint appears without anyone touching a mapping file.

The usual reason to switch is that SharePoint is the whole job. Running a data-integration platform to reach a handful of lists is a lot of machinery; the table below covers the desktop app and the reporting joins as well.

Can Airbyte feed a SharePoint RAG pipeline?

“We followed the SharePoint RAG tutorial and it indexed a file, but our knowledge lives in lists.” Airbyte’s tutorial puts a CSV file in a SharePoint folder, reads it through the SharePoint source and writes it to Milvus on Zilliz. The Milvus destination does the AI work: it splits records into chunks, embeds them with an OpenAI model and indexes them, and LangChain then retrieves from the collection. For document libraries that pattern works, and it is easy to extend to other vector stores Airbyte writes to.

Airbyte also runs Airbyte Agents, a context layer that gives agents live API access through an MCP endpoint. Its agent connector catalogue lists no SharePoint or Microsoft Lists connector (checked 2026-10-23), and Airbyte says those connectors are not built for replication.

SPSync covers the list side. It keeps one SQL copy of your lists and library metadata, read once by delta query and never written back to SharePoint. An agent queries that copy through a database MCP server you choose, or through spsync mcp serve, SPSync’s own read-only MCP server, under a read-only database login, so the agent holds no Microsoft Graph permission. With embeddings turned on, SPSync generates vectors on each sync for pgvector and Azure AI Search, the only two destinations that support it today. An edited item is re-embedded, and a deleted one leaves the Azure AI Search index or is flagged in pgvector so retrieval can filter it out. SPSync ships no agent of its own and copies metadata, not file text. SharePoint item permissions are not carried into the replica, so database roles and row-level security decide what each agent can read.

Moving the SharePoint lists across

1. Leave your Airbyte connections for other sources running; SPSync does not touch them.
2. In SPSync, sign in with delegated access or an app registration limited by Sites.Selected, and choose the site and lists Airbyte was pulling.
3. Point SPSync at the destination database and let the first full load finish; delta runs follow on the schedule you pick.
4. Check row counts and a few lookup joins against the old Airbyte tables, then move downstream reports and pipelines to the SPSync tables.
5. When everything agrees, disable the SharePoint connection in Airbyte so you stop paying for it.

The rest of your Airbyte setup stays as it is. This replaces one connector.

SPSync as an Airbyte alternative for SharePoint lists

Teams that replace Airbyte for SharePoint keep their other Airbyte connections and move only the lists. SPSync handles the SharePoint side as a desktop app or a command-line job, writes a schema shaped by each list's column types, and leaves the warehouse loads from other sources where they are.

Other alternatives

• Fivetran: a managed Microsoft Lists connector billed on Monthly Active Rows, for teams that want nothing to host.
• Estuary: streams files from SharePoint document libraries into a pipeline; list items are outside its SharePoint connector.
• CData Sync: one licence across many sources, with SharePoint read through a generic driver that you map yourself.
• Peliqan: a hosted platform with its own warehouse and BI connections, for teams that want lists and dashboards in one subscription.