A music company pays royalties to a roster of artists, but the data behind those payments lives in half a dozen disconnected places: spreadsheets, email attachments, cloud drives, accounting software, signed contracts. Answering "what does this artist's royalty picture look like?" meant hunting through all of them by hand, every time.
The system designed here fixes that with an always-on AI agent: drop a file in and it gets read, matched to the right artist, and filed automatically. Staff ask for any artist by name and get the full picture in one view (financials, contracts, even audio and artwork) instead of checking five systems.
The non-negotiable was trust. These numbers decide what artists get paid, so the design draws a hard line: the AI reads, organises and answers questions, but the actual money arithmetic runs through ordinary, auditable code, and every figure in every report links back to the exact file it came from. If two sources disagree, the system flags it for a human instead of quietly picking one.
The AI organises and answers; auditable code does the arithmetic. Nothing is estimated.