AI-Assisted Due Diligence for Catalog Acquisitions

In short: The Noctil Platform speeds up due diligence on music catalog acquisitions. Its ProcessEngine workflow engine coordinates AI models to read scanned agreements, pull out rights, ownership, territory and term data into a structured form, and link each agreement to its assets and royalty history. Buyers and their advisors cut more than 80% of the manual review work before closing. They also receive data ready for post-acquisition operations with less than 30% of the usual integration effort.

The Situation

When a music catalog changes hands, the buyer or its advisors receive a data room. It typically holds executed contracts, spreadsheets of asset metadata, and historical royalty statements.

Before committing capital, the buyer has to judge four things: the quality of the catalog, its value, its future revenue potential, and any legal exposure. That means detailed due diligence, matching every contract to the assets it covers and to the revenue those assets have actually earned.

The Problem

The material in a data room is rarely ready to analyze:

• Scanned, unstructured agreements. Most contracts arrive as copies or scans. Each one follows its own format and structure.
• Critical terms scattered across documents. Rights, territory, term, ownership and copyright registrations sit in different clauses. Asset lists are often buried in appendices, schedules or side letters.
• Metadata disconnected from contracts. Asset metadata comes separately, across many spreadsheets and PDFs, with nothing linking it to the agreements that govern it.
• Slow, manual review. Reading, extracting and cross-referencing all of this by hand takes legal and operations teams weeks.
• Hidden risk. Under deadline pressure, important terms such as reversions, restrictions and ownership gaps can be missed. That exposes the buyer to legal and financial loss, and puts the reputation of the investment firm or broker behind the deal at risk.
• Duplicated effort after closing. Once the deal closes, the operations team often repeats the same exercise to load the catalog into its own systems.

The Solution

The Noctil Platform is music and media metadata infrastructure built for AI-assisted workflows. ProcessEngine, its workflow engine, coordinates the AI models and automation that do most of the work.

1. Read scanned agreements. AI models read scanned and PDF contracts, including appendices and related documents.
2. Convert terms into structured data. Rights, ownership, territory, term, copyright registrations and other legally relevant clauses are mapped consistently into a structured agreement record.
3. Ingest asset metadata. Spreadsheets and PDF schedules of works and recordings are loaded and normalized.
4. Link agreements to assets. ProcessEngine links each agreement to the assets it governs. It infers the connections from the agreement itself and from supplementary documents.
5. Connect revenue history. Historical statements are tied back to the linked assets, so buyers can see what each asset has earned under which agreement.
6. Flag gaps and risks. Missing ownership, conflicting terms and unlinked assets are surfaced for review before they become post-close problems.

Model-flexible by design. ProcessEngine works with leading AI models, including Google Gemini, Anthropic Claude, OpenAI and DeepSeek. It also works with Noctil’s own models, trained for specialized tasks such as music and media agreement mapping. Each task goes to the model best suited to it.

Accuracy that improves over time. Every workflow runs against evaluations (evals) that measure extraction accuracy. Continuous feedback loops then refine the prompts, models and rules, so results become more accurate with each catalog processed.

The Result

• 80%+ less manual work in pre-acquisition review and due diligence
• Under 30% of the usual effort to make the data ready for post-acquisition operations, because the diligence output feeds straight into operational systems instead of being rebuilt by hand
• 95%+ of material contract terms extracted and linked, cutting legal and financial exposure before closing
• A single linked view of agreements, assets and historical revenue for valuation and investment committee decisions