Every organization is sitting on a goldmine of contracts (NDAs, service agreements, vendor contracts, employment letters) and most of them are locked inside PDFs, inaccessible to search, analysis, or automation. Legal and operations teams spend hours manually hunting for expiration dates, renewal terms, and notice periods. The process is slow, error-prone, and doesn't scale.
At OneSix, we saw Snowflake's newly released Cortex Code (aka CoCo) as an opportunity to solve this problem, fast. What followed was a proof of what's now possible when a next-generation AI coding agent meets a high-value enterprise use case.
The business cost of poor contract visibility is real: missed renewal windows, overlooked obligations, compliance gaps, and strained vendor relationships. For most organizations, the bottleneck isn't a lack of data; it's that the data is trapped in an unstructured format that traditional systems can't act on. This is precisely the kind of problem that Cortex Code was built to accelerate.
OneSix developed a fully automated Contract Intelligence solution entirely within Snowflake's ecosystem — from PDF ingestion through AI-powered extraction, data transformation, and a polished interactive interface. The result: a system that turns a static PDF into structured, searchable, and shareable intelligence in minutes. Key capabilities include:
All of this runs entirely within Snowflake's governed infrastructure, meaning sensitive legal data never leaves your controlled environment.
What made this build remarkable wasn't just what we built — it was how fast we built it. Cortex Code acted as an intelligent development partner that understood our Snowflake environment, not just generic code patterns.
Unlike a general-purpose AI, Cortex Code has native knowledge of your Snowflake environment — your schemas, roles, and data. When we kicked off the intake build, it proactively recommended a Role-Based Access Control framework, ensuring governance was baked in from day one rather than bolted on later.
Cortex Code doesn't act autonomously — it collaborates. Before executing any action, whether creating a table, changing a role, or running a query, it presents its plan for review and approval. This four-step loop of prompt, clarify, approve, and validate kept our team in full control while dramatically reducing the back-and-forth typically required during complex builds.
Throughout the build, Cortex Code diagnosed and resolved issues that would have cost a developer hours of debugging — from handling special characters breaking file ingestion, to refreshing stage metadata, to optimizing stored procedure logic. Each blocker became a brief conversation rather than a context switch into documentation.
The solution is a multi-layered data pipeline, all orchestrated within Snowflake:
Snowflake Stages serve as the PDF landing zone. Snowflake Streams monitor for new files and automatically trigger a Task to begin processing.
Snowflake Cortex AI services handle OCR, layout detection, and legal clause identification, converting unstructured PDFs into machine-readable, structured data.
A dbt pipeline (authored by Cortex Code) stages, cleans, and transforms raw extraction output into analytics-ready data marts.
A native Streamlit application provides contract browsing, clause highlighting, natural language queries, and PDF export, all within Snowflake.
Cortex Code generated technical documentation throughout the build, exportable to Excel and Lucidchart-compatible formats for lineage tracking, compliance, and future maintenance.
The Contract Intelligence solution is one example of a broader shift in what's possible. Cortex Code lowers the barrier between a business problem and a production-ready data solution dramatically. Teams no longer need to choose between speed and quality, or between innovation and governance.
For organizations looking to unlock value from unstructured data, reduce manual legal review, or simply move faster on data initiatives, this is a signal worth paying attention to.
