A health check that quotes your own documents

A snapshot of a project that is running: the plan or tracker (Excel or CSV works best), the last few sets of meeting notes, the latest status report, a risk or issue log. Up to five files. Different kinds of document matter — the most useful findings come from comparing them with each other.

Every finding has to quote your documents word for word, and the quote is checked against the text you uploaded — a finding whose quote cannot be found is dropped. The ones that matter most are then put to a second, independent check that can confirm, reword or reject them. Numbers such as overdue tasks are counted from your plan, not estimated. Your report shows each quote and the document it came from.

No. A baseline is computed from the verified findings and the counts from your plan, and the model's own rating is held close to it. The report lists what drives the score, so you can disagree with a specific line rather than with a number.

They are used to produce your report, kept for 30 days so the invitation in your email can turn them into a Soriela workspace, and then deleted. They are not used to train models. Project documents name people — upload only what you are allowed to share, and ask us if you want them deleted sooner.

The same reading, continuously, across the tools your team already works in — planning tools, chat, mail, meetings — with the health score, risks and drafted status updates kept current. The email you receive includes an invitation that opens this same project in a Soriela workspace.