Pattern: The Extraction Pattern

The Extraction Pattern is the systematic process by which human expertise is documented, modeled, and transferred to non-human systems. It operates through three phases: knowledge capture, pattern modeling, and operational transfer. The process is rarely announced. It is usually disguised as "knowledge management," "process improvement," or "digital transformation."

The defining characteristic of the Extraction Pattern is that the humans whose expertise is being extracted are typically the ones doing the extracting. They are asked to document their workflows, train the systems, and validate the output — building the infrastructure that will replace them.

The extraction process follows a consistent operational sequence.

Phase 1 — Documentation Request: The organization requests that experts document their processes, decision frameworks, and institutional knowledge. The stated reason is "business continuity" or "knowledge management." The actual function is creating training data.

Phase 2 — System Training: The documented knowledge is used to train automated systems — initially positioned as "tools to help you work faster." Experts interact with the tools, providing feedback that improves accuracy. Each correction is a training signal.

Phase 3 — Operational Transfer: The system reaches adequate performance on routine tasks. The expert's role narrows to handling exceptions, quality review, and edge cases. Headcount reduction follows within 6-18 months.

Phase 4 — Residual Compression: The remaining expert roles — exception handling, quality assurance — are themselves subjected to extraction as the system handles progressively more complex cases. The expert population compresses to a skeleton crew.

Call center offshoring (2000-2005): American call center expertise was extracted through script documentation, call recording analysis, and process standardization. The knowledge was transferred to overseas workers, then to IVR systems, and now to AI chatbots. Each transfer compressed the previous workforce.

Legal discovery (2005-2015): Document review — once performed by junior attorneys at $200-400/hour — was first transferred to contract attorneys ($30-50/hour), then to offshore review teams ($15-25/hour), then to AI-assisted review systems ($1-3/document). Each transition extracted the pattern recognition expertise of the previous tier.

Radiology (2018-present): Radiologists trained AI systems by labeling images, identifying anomalies, and providing diagnostic feedback. The systems now perform initial screening at accuracy levels that match or exceed junior radiologists for common conditions. The extraction was framed as "building better tools."

"Centers of Excellence" are being created to centralize expertise — a consistent precursor to extraction. When knowledge moves from distributed individual practice to centralized documented practice, it becomes extractable.

"Knowledge transfer" initiatives are accelerating across professional services, healthcare administration, and financial analysis. The language of collaboration masks the mechanics of extraction.

Organizations that have completed Phase 2 (system training) are entering Phase 3 (operational transfer) across legal operations, customer success, and technical support. The timeline from Phase 2 to Phase 3 has compressed from 18-24 months to 6-12 months.

You've been asked to "document your workflows" for reasons that seem reasonable but feel urgent.

New tools have appeared that mirror your decision-making process — tools you may have helped train through feedback or corrections.

Your organization has created a "Center of Excellence" or "Knowledge Hub" that centralizes what used to be distributed expertise.

Junior staff can now replicate 70-80% of your output using tools that encode your patterns.

Your role has shifted from doing the work to reviewing the work that systems produce.

The number of exceptions you handle — the work only you can do — is shrinking each quarter.