Agora TIS — Task Intelligence System
See what your team is working on — and change it by asking, in plain English.
Live internally; scoped to one project, widening next
- 92ms — Warm load of a person's standup view, down from 7.9 seconds — the difference between a tool people open and one they avoid
- 0 — Databases of its own — it reads Plane and GitLab live, so it can never disagree with the system of record
- 6 — Kinds of change the assistant can make in Plane, every one previewed and held until a human clicks Confirm
Work is tracked properly, and still nobody can see it
The information is all in the tracker. Assembling what is my team actually doing right now still means clicking through a dozen views.
So status gets asked for in chat, answered from memory, and is stale by the time anyone reads it. The people who most need the picture — leads, whoever runs standup — are usually the least willing to live inside the tool. And every dashboard built to fix that becomes a second source of truth, which then starts quietly disagreeing with the tracker.
One screen, and plain English
- The morning view. Projects, ticket counts, a card per person with their workload and what they last moved. The picture you would otherwise assemble by hand.
- Ask it. “What is Karl working on?” · “Who has the most on right now?” It reads the team’s real tickets to answer, not a summary someone wrote last week.
- Tell it — it asks first. Comment, change status, reassign, set priority or a due date, create a ticket. Each one is shown as a preview with a Confirm button, and nothing moves until a person clicks.
- Permissions the tracker already knows. Anyone on an approved domain can read. Only members of that project can change anything, matched by email and effective immediately.
- The brief sits with the ticket. Open a ticket and its full brief is there, with a link to the source — so the why is next to the what.
It proposes, you decide
The assistant never acts on its own. Ambiguity produces a question rather than a guess: ask it to comment on “the connector ticket” when three of them match and it asks which one you meant. A preview is a proposal, and Cancel discards it.
Reading and writing are separate rights, and both are checked on the server for every action rather than in the browser.
Nothing to keep in step
It has no database. Every answer is read live from the tracker, and every change it makes is written back there, where the rest of the team already looks.
That is the design decision the whole thing rests on. A reporting layer with its own copy of the data will eventually disagree with the system of record, and then someone has to reconcile the two. This one cannot drift, because it has nothing of its own to drift with.
Where it fits
- Teams whose answer to “where are we?” takes more clicking than it should.
- Leads and stakeholders who need to see delivery without being trained on the tracker.
- Anyone wary of dashboards that quietly become a second source of truth.
- Teams curious about an AI assistant that acts on real systems, with a person approving every write.
See it on a board you know
The quickest way to judge it is to point it at a project you know well and ask something you already know the answer to.