Progress and the reading estimate
Follow the app as it reads a case, understand the stages and the time estimate, measure your computer's speed, and deal with files that failed.
#The Progress screen
Open Progress in the left column. The page title is the case's name, and the line under it says where things stand:
| Status line | Meaning |
|---|---|
| Not started yet. | Nothing has been read. Click Start reading. |
| Reading files. About 25 minutes left | The app is reading; the time left updates as files finish. |
| Paused. N files are not finished. | Reading stopped with work left. Click Resume. |
| Done. Nothing is left to read. | Every file has been through every stage. |
The buttons at the top:
| Button | What it does |
|---|---|
| Estimate reading time | Opens the reading-time estimate (below). |
| Start reading / Resume | Starts or continues reading. |
| Pause | Stops after the files in progress. Nothing finished is lost. |
| Rescan folder | Looks for new, changed and removed files in the case's folders, then reads what's new. |
| Add files or folders | Adds another folder or files to the case. See Adding files and folders. |
#Searchable now
Near the top, Searchable now counts the files you can already search. Search, the document viewer and your AI agent all work on whatever has been read so far, so you can start long before the case is finished.
#Reading now
While the app is reading, a Reading now panel shows:
- Elapsed, Time left, Total time and Finishes around (a clock time).
- Files done, as a count and a percentage, with how many could not be read.
- Speed: files made searchable a minute.
- A bar of files Done, Being read, Could not be read and Waiting.
- By kind of file: PDFs, pictures, e-mail, recordings and so on, each with done, reading and time left.
Early in a big case the time left is marked rough, until files are read: it is based on typical times until the app has seen how fast your computer reads this case.
#Stages
Every file goes through these stages in order:
| Stage | What happens |
|---|---|
| Unpack archives | ZIP files are opened and their contents listed. (Shows No archives if there are none.) |
| Read | Text is taken out; scans are read with OCR; recordings are transcribed. |
| Find personal details | Names, accounts, addresses and similar details are detected. |
| Link names | Every spelling of a name is tied together, so search finds them all. |
| Make searchable | The text is indexed for search by words and by meaning. |
Each file becomes searchable as soon as it is through the last stage.
#When a stage waits for a model
If a model hasn't been downloaded yet, the files that need it wait and the rest carry on. Progress says so in plain words, with a button to fix it:
| Message | Button |
|---|---|
| Recordings wait for the speech model. | Download the speech model |
| Finding personal details waits for the name model. | Download the name model |
| Making files searchable waits for the search model. | Download the search model |
Once the model is installed, reading picks the waiting files up by itself. You can leave the screen while it downloads.
#Time
The Time section shows the time left, split by kind of file, with a note on what the estimate is based on:
- estimated from the files read so far (the best basis),
- estimated from earlier cases on this computer, or
- a rough estimate from typical times, until some files are read.
Where the time went breaks down each reading run: how long it took, how many reading processes it used, which kinds of file took the time, how many scanned pages were read with OCR (usually most of the time) and the slowest files.
#The reading-time estimate
Click Estimate reading time before you start a big case (or any time). The Reading-time estimate panel looks at the case's folders without reading anything (only the files' names, sizes and first few bytes) and tells you:
- how many files there are to read, and how long they'll take on this computer, on how many reading processes;
- a table by Kind of file: how many there are, how many are Read already, and the Estimated time for each (or Waits for a model);
- notes such as how many files are ZIP archives, how many won't be read, and how many couldn't be opened.
Click Look again to refresh it, or Close.
#The speed score
The panel also shows this computer's Speed score: how many five-page scanned PDFs it reads in 10 minutes. Click Measure this computer (later Measure again) to measure it. The test reads a made-up five-page scan on every reading process at once; it takes about 20 seconds and uses every processor core, so it waits until reading has finished or been paused. Nothing from any case is used.
#Files
The Files table counts every file by how far it got:
| Row | Meaning |
|---|---|
| Indexed | Read and indexed, or opened (archives, production pages). |
| Partly read | Some parts could not be read. |
| Waiting | Not finished yet. |
| Duplicates | Exact copies of another file, read once. |
| Production load files | Used to assemble a production. |
| Not read | Types and links the app does not read. |
| Could not be read | Empty, damaged, password-protected or refused by the system. |
| Failed | See the failures below. |
| Missing from the folder | Gone from the case folder; their work is kept. |
#Failures
A file that fails is retried once automatically. If it fails again it is listed under Failures with the stage and What happened, in plain words. For example:
- The file changed after it was listed. Retrying reads the new version.
- Another program has the file open.
- A reading process stopped, often from lack of memory.
- The disk is full.
Fix the cause if there is one (close the other program, free some disk space), then click Retry on a row, or Retry all N.
#If live updates stop
If Progress says Live updates paused … Reconnecting…, the app is reconnecting to its engine by itself. If it says Engine access refused, restart the app. If reading seems stuck with no change for a long time, see Troubleshooting.