Choosing an AI Vendor: The Five AI Archetypes

Legal ops needs to know the five AI archetypes  to choose an AI vendor that will improve workflow and performance. Determining successful legal AI adoption depends on the quality and governance of the data supporting the program, writes Sinead Kenny of Brightflag.

The term “AI for legal ops” means applying AI to running the department, which includes invoice review, spend management, workflow automation, outside counsel oversight, and reporting. These capabilities rely on structured operational data, enabling AI to automate routine processes, recognize patterns, and identify exceptions that require human judgment. A critical first step is transforming unstructured information into structured metadata so AI can consistently interpret invoices, billing records, and other operational information.

To help evaluate AI vendors and technology, Kenny outlines five AI archetypes. First are systems of record, which serve as the authoritative source for governed operational data and business rules. Next, data warehouses consolidate information from multiple systems to support enterprise-wide analysis, while AI workspaces provide natural-language interfaces that allow users to query operational data and generate insights. Also, AI agents carry out defined tasks, such as reviewing invoices or initiating workflows. Finally, AI enrichment tools convert unstructured information into structured data, creating the foundation that enables the other archetypes to operate effectively.

Kenny argues that the data layer is more important than the AI layer because reliable outcomes depend on governed, structured information. Advanced AI operating on incomplete or ungoverned data can produce inaccurate results, creating significant operational and compliance risks. Strong governance, auditability, and clearly defined business rules are essential prerequisites for trustworthy automation.

In choosing an AI vendor, Kenny recommends focusing on architectural fundamentals rather than marketing claims. Legal ops teams should determine how unstructured data is transformed into structured information, whether AI is governed by a system of record, and which of the five AI archetypes the platform actually includes.