Power Tables

Pay-as-you-go on rows stored: $1.20 per 1,000 rows per year for the first 1 million rows, $0.36 per 1,000 to 10 million, and $0.24 per 1,000 above that. There is no per-seat charge and no monthly floor. The app is available to every tenant with no license decision — it simply stays off until an admin enables it.

There are three app-level tiers: administrators create tables and define columns, contributors add and edit rows, and everyone else with app access can open tables, filter, sort, and search — viewing is not an admin privilege. On top of that, each table is shared individually: leave it open to everyone with app access, or mark it restricted and name its audience by granting specific people or recipient groups viewer, contributor, or manager rights. A grant on a restricted table can raise someone to contributor for that one table without giving them the app-wide tier.

Import accepts CSV files up to 500 MB and Excel (XLSX) workbooks up to 100 MB — for a workbook, the first sheet imports — through a two-step flow where you map each file column to a table column, create it as a new text column, or skip it before anything lands. Export goes the other way as CSV or Excel, capturing the grid's current view (filters, search, sort, and visible columns) as a background task with a download button when it's ready; Excel export is capped at 100,000 rows, CSV has no cap.

Both directions can also run unattended. Each table can have one scheduled import — a nightly pull in replace mode (the source is the system of record: rows are cleared and re-imported) or append mode (every run adds rows, for accumulating logs) — and one scheduled export, which writes the file to a path on an SFTP destination. A Run now button triggers either on demand, and the last run's status, including failures, shows on the table page. A scheduled import's source can be a public https URL with no setup at all, or a protected source through a connection an admin configures once at /admin/integrations: a Data Feed connection sends its credential (a bearer token, basic auth, or a custom header) with the nightly HTTPS pull, and an SFTP connection fetches a remote file path instead of a URL. For programmatic access there is also a REST API.

Yes, and that is the point of counter columns. Mark any number column as a counter and it stops being a typed field — it renders − / + buttons in the desktop grid, in the row drawer, and on a phone. A tap sends the change rather than the answer, and the server applies it under a row lock, so if a shelf holds 12 and two people each tap − at the same moment the result is 10, never 11. Counts floor at zero, and crossing zero can fire an automation so someone is told to reorder.

Each adjustment can carry a short note — "used in OR 2" — saved with the change. That rides the same audit trail as every other single-row edit: each add, edit, and delete through the row drawer, bulk edit, or the REST API is logged per field, showing who changed it and the old value to the new, kept for 180 days and shown in the row's own History panel. Imports and bulk inserts are deliberately not logged row-by-row; the import task record is the audit trail there.

Yes — that's what Mini Apps are, and they're created in App Builder rather than here. An admin describes what they need in plain language and gets a real app at its own address, with its own name, icon, and place in the sidebar. Underneath, it's Power Tables: the app's data is a table, its tabs are saved views, its behavior is automation rules, and its dashboard is a Live Board. Because it's a real app, it carries its own permissions, so a frontline user can be given a single narrow door — report a count for one item — without being able to browse or edit the table behind it. Mini Apps you create stay private to your tenant; they never appear in the public app directory, and their rows bill on the same Power Tables row meter rather than a second charge.

Yes. The REST API under /api/v1/power-tables lists tables, returns a table's structure (columns and views), reads rows with the same filter grammar and keyset pagination as the grid, runs filtered aggregates (count, sum, avg, min, max), and — for contributor users and above — ingests rows one at a time through the same casting, quota, and validation path as the row drawer. Access is by API token with least-privilege scopes: read:power_tables for reads, write:power_tables plus Power Tables contributor rights for writes.

No — that is the point of the architecture. Each table is a real Postgres table, so filters and sorts are SQL with real indexes, pagination is keyset-based (page one thousand is as fast as page one), and row counts are fast estimates rather than blocking scans. You mark the columns you sort and filter on for indexing, and index builds run in the background with tracked progress.

A saved view captures a filter, a sort, and a set of visible columns — authored straight from the grid, and every table starts with an "All rows" view. Live Boards bind to views, not raw tables, so the view is where you define exactly which rows and ordering a leaderboard or stat tile reflects. Refine the view and every board on it updates.

Yes — each table can carry automation rules. A rule fires either on a row event (row created, row updated, a named field changed, row deleted) or on a schedule (every N minutes down to a 15-minute floor, daily, or weekly). Its actions can update fields, notify a user, send an email, request an approval, create a task, call a webhook, lock a row, archive it, delete it, or run an AI action — fill a column, post a summary digest, raise an anomaly alert. Approvals carry per-outcome follow-on actions, so an approved row and a rejected row can take different paths. One guard is deliberate and worth knowing: a scheduled rule can never message per matching row, so a rule matching 10,000 rows updates them silently or reports once as a digest rather than sending 10,000 notifications. Every rule keeps a run history and has a Run now button. Separately, a view watch can alert when a saved view's count or aggregate crosses a threshold.