Roadmap
Where this is going.
Three lists, honestly labelled. Available — in production today. In development — being built now. Planned — decided, not started; order may change. If an item on this page matters to your decision, tell us — it’s how we prioritise.
In production
Running in customer applications today.
Object-level access control
Permissions declared over your model’s own relations, enforced on every read, search and change — with the path along which access was granted.
Model compatibility guarantees
A model written for one release keeps loading on later ones. Format changes hand back your repository rewritten, with a report.
AI document processing
Classification, extraction and invoice reading, with token usage and cost recorded per call.
Company register integrations
KvK, Companies House and OpenCorporates lookups, organisation trees and corporate relations.
Live analytics replication
A continuous copy of selected data in your own BigQuery dataset for dashboards.
Being built now
Designed, and under construction.
Models that check themselves
Verifying a model in a disposable environment: unresolved references and missing translations reported per language before anyone opens the application.
Decided, not started
In priority order. Order may change.
Auditing and circuit breakers
Comprehensive access auditing, and protection that detects and halts anomalous bulk extraction.
Documents we cannot read
File storage in a location you control, with your own backups. The platform holds keys, never content.
Exit tooling
Documented export of data, model and documents in open formats — portability you can demonstrate, not just be promised.
An agent-ready register
A governed tool surface, generated from your model, that any AI agent or copilot can use — within the same permissions as your people, with writes gated per operation.
Your model, in plain language
A generated, readable description of your whole system, for readers without database knowledge. It can never go stale.
Hosted test environments
A persistent place to review model changes before they reach production.
Pre-action validation
Scripts that decide in advance whether an action is permitted at all.