TrmTrk Docs

Appearance

Features/Uploading Contracts

Uploading Contracts ​

Term Tracker reads PDF and DOCX (Word) files. Drop them onto the Documents page and the rest happens automatically: classification, queuing, analysis.

The Drop Zone ​

When a project is empty, the entire Documents page is one large drop zone. The whole workspace prompts you to drag files in.

Once the project has at least one contract, the drop zone compresses into a slim upload bar at the top of the Documents list. Same functionality; it simply gets out of the way of the contracts you are already working with.

Multi-File Upload ​

You can drop ten files at once, or a hundred. Every file uploads in parallel and each one queues its own analysis independently.

It is fine to grab a full folder from your data room and drop it onto the page.

Auto-Classification ​

As each file uploads, the AI runs a quick structure pass and detects its document type (MSA, SOW, NDA, Lease, or whatever types the project's template defines). You do not pick the type yourself.

The detected type shows up as an editable badge on the contract's detail page. If the AI gets it wrong (it happens with shared templates or unusual headers), click the badge to correct it.

Status Badges ​

Every contract in the Documents list carries a status. The left border of each row color-codes it so you can scan the list at a glance:

  • Uploading: file is still transferring to S3.
  • Ready: uploaded, not yet analyzed.
  • Analyzing: AI extraction is running.
  • Complete: analysis finished, matrix cells populated.
  • Error: something went wrong; click the row for details.

How Analysis Runs ​

Every analysis queues to a background worker; there is no file-size threshold and no inline path. You do not have to keep the tab open. Close it, come back later; the results will be waiting, and you will get an email when the job finishes.

→ For more detail, see Running AI Analysis.

File Size & Naming Tips ​

TIP

A clear filename makes the Documents list much easier to scan. The AI does not care what the file is called (it reads the contents), but 2024-Q3-MSA-AcmeCorp.pdf browses better than scan_003.pdf.

TIP

A small scanned or photographed PDF (up to 20 pages and 20 MB) is read directly by the model as page images, so it extracts without a separate OCR step. Larger scans fall back to the text pipeline: run those through an OCR tool (Adobe Acrobat's Scan & OCR, or any dedicated utility) and split them under the size limit before uploading. Text-based PDFs and DOCX files always extract directly. If a large scanned contract comes back with an extraction error, check that it is text-searchable first.

→ For more detail, see Running AI Analysis. → For more detail, see The Analysis Matrix.