Pydio Cells vs Teable

Teable ships numbered releases nearly every day, and this window is dominated by administration rather than end-user features. A CLI workflow now exports, edits, previews, compares and applies declarative permission matrices with safeguards against accidental lockouts; audit logs record specific record IDs, the API endpoint behind each action, and space and base lifecycle events; admins can cap AI concurrency per space and issue single-use password resets. The user-facing changes are comparatively small — mobile controls on shared pages, in-app invite notifications, Kanban dialog affordances.

The work is assembling what a self-hosted, multi-tenant deployment needs before an IT function will approve it: declarative permissions, provenance in the audit trail, blast-radius limits on AI work, and tighter sandbox rules for agents that install their own dependencies. AI is being handled as a workload to govern — five concurrent tasks per space by default, queued fairly across field generation and automations — rather than as a feature to advertise. Steady reliability work on computed fields, concurrent SQL and table loading runs underneath all of it.

Expect the permission-matrix workflow to extend across more of the admin surface, and sandbox and concurrency controls to tighten further as agent activity inside bases grows. Nothing in these releases points to a new end-user capability landing soon; the near-term direction reads as enterprise readiness.