Your operational data, inside the room where the work happens
Not a database in another tab. A table belongs to an agentic room, so the people in the conversation, the tasks, the meetings and the approvals are the same people, tasks, meetings and approvals — and the AI project manager in that room can read and write every row.
A spreadsheet knows numbers. A table in a room knows your company.
Because the table lives where the work is discussed, a column can hold a person, an approval or a link to another table — and mean the same thing everywhere else in the office.
People are people
A person column holds a real member of your office — with their reporting head, department and title attached. That is what lets a process route to “the requester’s manager”.
Approvals are approvals
An approval column is a real approval: it appears in the approver’s queue, sends the email, collects several decisions — and is still a value you can filter and count.
Every row can hold a conversation
Discussion about one record stays on that record, where the next person to open it will find it — not in a thread nobody can search.
The agent is a first-class writer
Your AI project manager creates rows, updates them and reads them back with the same permissions a person has. No integration, no connector, no sync.
Rows start things
A row entering a state can start an automation or open a workflow case — the table is the trigger surface, not a dead end.
Tables relate to each other
Link records across tables in the room, roll values up, and compute totals that stay correct as the underlying rows change.
One set of records, four ways to work with it
A view is a lens, not a copy. Change a record anywhere and every view — and every person watching — has it immediately.
Grid
The dense, keyboard-fast default: filter, sort, group, hide columns, set row height and colour rules, then save it for the team.
Board
Grouped by any select or person column. Drag a card to move it — which is also the fastest way to move a workflow case along.
Calendar
Any date column laid out by week. Drag an item to reschedule it; the record changes, not a copy of it.
Gallery
Cards with attachments shown large — for candidates, products, properties and anything people recognise by sight.
Twenty-two column types, and the computed ones stay correct
Formulas and rollups are stored, and recomputed down the chain the moment something they depend on changes — so a total is never a stale number waiting to embarrass somebody.
Formulas that hold the rules
Total commitment, discount percentage, days requested, days overdue — the arithmetic a process branches on lives in the table, not in somebody’s head.
Rollups across linked records
Sum the invoices against a contract, count the open tickets for a customer, take the latest date from a set of child rows.
Attachments that count
Receipts, contracts, photographs — stored against the record, counted against your office storage, and visible in the gallery view.
Permissions people can actually explain
The room owner decides a table exists
Creating and reshaping tables is an owner’s job, so the room’s data model does not drift.
Editors work, members contribute
Editors can be given the schema, or only the rows. A member can be limited to the records they created — enforced on every write, not just hidden in the interface.
Sharing is explicit
A view can be shared as read-only or as add-a-row, with a link that stops working the moment you say so.
Nobody gets data by accident
Taking part in a process through a form never grants access to the table behind it.
Every change is recorded, and every change can be undone
History is the audit trail
Who changed what, when, and from what — the same record serves both the undo button and the auditor.
A 200-cell paste is one undo
Bulk changes undo as a single step, and undoing appends to the history rather than erasing it.
Two people, no silent overwrite
If somebody edited a cell while you were in it, you are asked which version to keep. A stale write is never merged quietly.
Deleted is not gone
Records and whole tables sit in the trash for thirty days, and restoring a table restores everything attached to it.
Built for a team working at the same time
You can see who else is here
Faces on the table, cursors in the cells, and the text somebody is typing before they commit it.
Comments where the decision was made
Discussion attached to the row, with mentions that reach people wherever they are in the office.
Import without a project
Bring a spreadsheet in as it is. If it would exceed the table’s limits you are told before anything is written, never halfway through.
Export anything, any time
Any view as CSV or Excel, or the whole table as a backup that can be imported again — to duplicate it, move it, or restore it.
Room-scale by design
Up to 10,000 records and 200 columns per table, 50 tables in a room — sized for how a team actually runs a process.
The ledger a process writes to
When a workflow runs, its cases land here as rows — so the process and the data are never two different systems.
Move the spreadsheet into the room
The tracker everybody copies, the sheet only one person maintains, the list that drives a process nobody documented. Bring one in and watch what happens when the agent, the approvals and the conversation are already there.
Explore the rest of the platform
Every part of Qik Office shares the same rooms, the same people and the same permissions — so the work one part produces is immediately usable by the next.