Morphik
A good fit for teams whose documents are full of diagrams, scans and tables that text-only RAG pipelines mangle. The page-level embedding approach is the draw, and you can self-host the core. Check the license and the company's direction before building a product on it.
Pros
- Embeds whole pages, so visual content is not lost to parsing
- Includes knowledge graphs, extraction, a research agent and MCP support
- Self-host or cloud, with RBAC and multi-tenancy built in
- Free tier and a public GitHub repository with active commits
Cons
- Core code is under the Business Source License, not an OSI open-source license
- Small page allowances on Free and Pro; overage is billed per page
- Accuracy and speed claims on the site come from the vendor's own benchmark