Make the handoff make sense.
Connect the format you have to the destination you need. These are implementation patterns—not preconnected accounts or one-click third-party integrations.
Spreadsheets → application records
Choose the data region, preserve string identifiers, and map columns to a destination contract. Keep parsing and business validation separate.
Explore spreadsheet importsWebsite content → editorial records
Use an authorized source, preserve headings and links, and keep extraction separate from publication review.
Explore content workflowsDocuments → a knowledge collection
Extract reviewable sections, attach source revisions, and evaluate content quality before indexing.
Explore knowledge importsImages → a media library
Connect originals, derivatives, descriptive metadata, and publishing status through stable asset identity.
Explore media workflowsKnowledge records → retrieval
Carry section context and permissions into derived chunks. Evaluate retrieval with representative questions and expected evidence.
Explore LLM importsPrompt definitions → an application
Preserve message roles and variables, then use a reviewed adapter and evaluate behavior before activation.
Explore prompt importsReview the boundary,
not just the connector.
A source system and a destination can agree on JSON while disagreeing on identity, permissions, dates, or update behavior. Capture those decisions before moving real records.
Compare source formats ↗Before connecting a destination
Identify the owner of the destination schema. Decide which fields can be updated, how missing values behave, and whether existing records are matched by a stable external identifier.
Before repeating the workflow
Document source revisions, credential handling in your own implementation, and a recovery path for uncertain outcomes. Test the exact configuration and permissions rather than inferring compatibility from a brand name.