Working papers, client records and engagement files, organised per client and reconstructable long after the engagement closes.
The test of a working paper file is whether a stranger can follow it two years later. Usually they cannot.
Engagement files are assembled under time pressure by people who all know the context, and that context leaves with them. These five stages put the structure in the file rather than in the team.
Clients send what they send: a folder uploaded directly, a year of schedules imported in bulk from a spreadsheet with progress shown as it runs, or material that already sits in a connected Google Drive, Dropbox or OneDrive. Email is indexed alongside documents, including shared organisation mailboxes, so the covering message and the attachment stop living in different places.
Every document carries a category, a document type and an owner, which is what makes a file navigable by a person who did not build it. Dates, reference numbers and parties are extracted on arrival, so the identifying detail is on the record rather than buried in the page.
Documents stay private to whoever added them until they are deliberately shared, and view, edit and approve are separate rights rather than one switch. A restricted record does not appear in another team’s search results, not even as a title, and each organisation’s records are stored fully separated from every other organisation’s.
Review runs as defined stages with named owners and deadlines, with reminders when a stage goes overdue and several named reviewers where more than one sign-off is required. A closed engagement can be reopened, so a late question reopens the case rather than starting a parallel one.
Every action sits in a tamper-evident history, which is the difference between a file you can explain and one you can only vouch for. On top of that you can ask for a summary of a document or a comparison between two of them, and every answer cites the document and the line it came from.
In each of these the document was there all along. What did not survive was the reasoning around it — who prepared it, what it was tested against, and why the version that was kept is the one that was kept.
A question lands on a file that closed last year, and the people who assembled it are on other clients or other firms. Reopening the closed case puts the new work on the same record rather than beside it.
Half of what made the file navigable was in one person’s head: which schedule mattered, why a number was accepted, who at the client sent what. Owners, document types and indexed email threads leave that in the file instead.
Representations and executed agreements arrive as scans or photographs, which is the point at which most files stop being searchable. Those pages are read on arrival in Nepali and English, so the signed copy is findable on what it says.
A client sends a revised schedule mid-engagement and the file quietly acquires two of everything. Draft, active and superseded are explicit statuses, and repeat copies are flagged on intake rather than found later.
Different stages, different jobs — from fieldwork through review and completion to the handover that follows.
Engagement
Pick up last year’s working papers with their structure and their correspondence still attached.
Review
Read the review approvals and a summary of the file rather than interviewing whoever prepared it.
Client management
Separate access per engagement, so a search cannot reach across clients even accidentally.
Fieldwork
Accept scans, spreadsheets and emailed attachments into the same engagement record without re-filing them.
Completion
Reopen without starting again, with the reopening itself recorded in the history.
Handover
Hand over the file rather than the context that otherwise lives in one person’s head.
Not unless they have been given access. Documents stay private to whoever added them until they are deliberately shared, and view, edit and approve are separate rights. Search results respect the same boundary, so a restricted record does not surface for a team without access — not even as a title in a result list.
By what is on the records rather than by asking the preparer. Every document carries a category, document type and owner, and the dates, reference numbers and parties are extracted on arrival. Beyond that, a reviewer can ask for a summary of a document or a comparison between two of them and get an answer that cites the document and the line.
Yes. Email is indexed alongside documents, including shared organisation mailboxes readable by administrators. That matters more than it sounds: the reasoning behind a working paper is usually in the thread that accompanied it, and that thread is normally the first thing to disappear.
A closed or rejected case can be reopened, so the later work attaches to the original record instead of starting a second one beside it. The reopening and everything after it goes into the same tamper-evident history as the original review.
Yes. Scanned and photographed pages are read automatically in Nepali and English, so a signed representation that arrived as a photograph is searchable on what it says rather than on the filename it was saved under. Handwriting is not supported.
Not yet. Retention rules by document type are being built and are not available today, as is access review reporting. What exists now is a tamper-evident history of every action and explicit draft, active and superseded status on every document, so nothing is lost while those are finished.