Content Portal · API & Integration

Everything the Interface Can Reach

A documented REST API covering ingest, delivery, and administration — under the same permission model as the interface, and structured the way a grounded retrieval layer needs.

Content in
Ingest is an API, not a manual upload screen with automation bolted on afterwards. Large files upload directly to storage through presigned URLs, and multi-document packages are unpacked and registered by the parser.
- Presigned upload for large files, bypassing the application server
- Package ingest with automated unpacking and registration
- Draft and publish transitions driven over the API
- Retry endpoints and failure statistics for unattended pipelines
Content out
Everything the interface can reach, an integration can reach — publications, versions, renditions, pages, and tables of contents are all addressable, under the same permission model as the UI.
- Publications, versions, and rendition endpoints
- Page-level retrieval, not just whole-document download
- Table-of-contents structure for navigation and chunking
- Search endpoints, with and without historical versions
Integration into your toolchain
The portal is designed to sit inside a wider system rather than to be the only place content lives. Organizations, departments, roles, and users are all API-managed, so provisioning can be driven from the system that already owns that data.
- Organization, department, role, and user management over the API
- Notification and access-request endpoints
- An events API accepting batched behavioral data, queued for asynchronous processing
- Token authentication, with the same authorization rules as the interface
Ready for grounded AI
The portal does not ship a chatbot, and this page will not pretend otherwise. What it does is hold your approved content in the shape a retrieval system needs: structured, page-addressable, version-aware, and permission-scoped at the source. That last property is the one most retrieval projects get wrong — an assistant built on a flat export will happily quote a superseded revision to a user who was never cleared to see it.
- Page-level addressing gives retrieval a natural chunk boundary with a citable URL
- Version and supersede relationships let an assistant answer from current content and say so
- Permission scoping is enforced at the source, so a retrieval layer inherits it instead of reimplementing it
- Shakewell builds the retrieval and assistant layer as an Applied AI engagement on top of this API