Knowella

Chat History — every run is a thread you can reopen

The thread list, the five stages every run moves through, and how to diagnose a workspace from its stage pills.

5 min read Real product screens
app.knowella.com Live walkthrough
Chat History is the workspace-wide list of runs. Every rating run KnowRFP has done appears here as a thread, stamped with the stage it reached, and any one of them can be reopened and steered with a message. The walkthrough reads the list before opening anything — because the pills across the list say something no individual thread does — and then goes inside one.
In this article

Five stages, in order

Every thread is somewhere on the same five-stage pipeline, and the pill on its card is how far it got. Knowing the order is what turns the list from a log into a status board.

  • 1 · IntakeFiles fetched and read.
  • 2 · Rating planDatasets grouped, blanks identified, method chosen.
  • 3 · RatedLanes priced.
  • 4 · Rate sheetsDeliverable workbook built.
  • 5 · DeliveredHanded to the customer.

What a thread card carries

A project chip, the project name, how long ago the run moved, and the stage pill. That is deliberately little: the card is for scanning, not for reading. Anything you want to know about how a run reached its stage lives inside the thread.

Read the pills before you open anything

Scroll the whole list and note the highest stage you can find. If a stage appears repeatedly and nothing goes past it, that stage is your bottleneck — and it is a property of the workspace rather than of any one run. The dashboard's Golden ready counter says the same thing from another direction, so the two screens are a cross-check on each other.

Diagnosing by stage

The distribution tells you what to do next, and the answer is different in each case.

  • Piling up at Rating planRuns are stalling on questions nobody answered. Go and answer them.
  • Piling up at RatedPrices exist but no deliverable was built. Finish the workbooks.
  • Piling up at Rate sheetsOutput is finished and unsent. The bottleneck is delivery, not rating.
  • Nothing at DeliveredNothing has ever gone out. Worth knowing before anyone reports on throughput.

Opening a thread

A card opens into the thread. The header identifies the project and the run, and a stepper across the top shows the five stages with the current one marked — the ones behind it complete, the ones ahead not reached. That stepper is the fastest answer to "where did this run actually get to", and it is the same on every thread, so you learn to read it once.

A thread can open empty, and that is by design

The page promises that every rating run is a resumable conversation, and a thread you open may well read "No messages yet". Both are true. The run itself happened in the Run panel, not here; a thread is a steering surface, not a transcript. Nothing was lost. What the thread gives you is somewhere to send the next instruction against a run that has already done its work.

Steering a finished run

The composer at the bottom takes plain English scoped to the whole run — regenerate a sheet, re-rate a dataset under different assumptions, exclude a file, push it to the next stage. This is the difference between a report and a workflow: to change one decision you reply to the run, rather than re-uploading the project.

How to use the page

Scan, then open. The list is for finding which stage is holding you up; the thread is for doing something about one run. Opening threads at random is how a long list stays long.

Ready to build your own?

Start free and ship your first app in minutes — or have us set it up with you.