A sheet-first operating record for structured-finance advisory.

The practice needed visibility across live matters, compliance gaps, document status, and fee events without forcing the team into a separate CRM. The real source of truth was already a spreadsheet - that was where the book of mandates was actually being maintained.

The design decision was to respect the behaviour that already worked, and to build everything else around it.

The system treats the spreadsheet as the operating record and builds advisory logic, document strategy, and product views around it, rather than pretending the team will maintain a separate database. The point is not to automate judgment away. It is to put the right facts in front of the person making the judgment.

The operating layer keeps live work, reference material, compliance records, and audit history in distinct surfaces so that no fact lives in two places at once. Advisory responses follow a fixed analytical shape so that the basis for any recommendation is legible at the time it is given, not reconstructed afterwards.

Document generation is deliberately conservative. Standard local outputs can be deterministic. Higher-risk or cross-border documents use template registries and bracket unknown legal facts rather than inventing entities, governing law, signatories, or security structures.

Local advisory work runs authority checks before any commitment-bearing language is treated as execution-ready. Cross-border work stays bounded by source documents: rights, permits, title, ownership, and lender-readiness are confirmed from primary sources before anything moves to an external recipient.

The system also distinguishes forecast economics from receivables. A possible fee event is not treated as a receivable until billing or collection is recorded. That distinction is basic accounting discipline, but it is exactly the kind of discipline that informal pipeline discussions tend to blur.

The next version would strengthen permissioning, build a cleaner read-only partner interface, and add structured review states for generated documents. The aim is not to replace legal judgment, but to make the state of the book and the next action harder to misunderstand.

The architecture is the public view. The private register, permissions, financial logic, and document-control details stay outside the portfolio.