A recent Sense Collective roundtable brought together legal operations and legal innovation leaders to compare the legal AI tools they are actively using, piloting, or evaluating. Members described how they are combining enterprise AI, legal-specific platforms, workflow tools, point solutions, and internal builds; how they are assessing security, defensibility, and eDiscovery readiness; and how they are trying to turn experimentation into measurable operational value. Below is a highlight of the discussion without attribution to any individual or organization.
1. A hybrid legal AI tech stack is emerging
The dominant pattern was a layered stack that combines enterprise AI for broad productivity, legal-specific platforms for content-grounded legal work, targeted point solutions for mature workflows, and internal builds where teams have the capability and governance to build safely.
This echoes the hybrid approach highlighted in Factor’s 2026 GenAI in Legal Benchmarking Report which found that 56% of respondents have purchased specialized legal AI tools, 49% have built an internal interface, and 34% have done both.
The most mature strategies sounded less like “choose one tool” and more like “decide what belongs at each layer.”
2. Workflow first, vendor second
Several Sense Collective members emphasized that the legal team needs to understand its own workflows before choosing tools. That includes mapping intake, identifying high-volume requests, understanding which data sources matter, and defining success criteria before a vendor selection process begins.
Similarly, 47% of respondents in Factor’s 2026 GenAI in Legal Benchmarking Report, named workflow redesign/orchestration as the lever that would accelerate AI impact in 2026.
A recurring pattern among The Sense Collective members was to use lightweight prototypes, enterprise tools, or general AI tools to clarify the workflow before buying a specialized platform.
3. Point solutions are winning where precision and workflow depth matter
The group did not treat general-purpose AI as a replacement for legal-specific tools. Purpose-built products came up repeatedly where quality, auditability, workflow depth, legal content, or integration with legal records matters.
Examples of use cases for point solutions included: surgical contract redlining, invoice review, legal intake and approvals, contract intelligence, corporate and securities workflows, marketing and consumer protection compliance, litigation drafting, outside counsel economics, and controlled litigation or investigation materials.
In the 2026 GenAI in Legal Benchmarking Report, respondents selected the following as the top in-scope areas for AI use cases: document review & summarization (72%), legal research/case law (63%), contract analysis & clause extraction (59%), and drafting & redlining (59%).
4. Governance and defensibility are central adoption blockers
Members were enthusiastic about powerful tools and agentic workflows, but governance concerns were a major limiter. The discussion repeatedly returned to tenant architecture, prompt movement, retention, privilege, legal hold, eDiscovery, local-file access, connectors, key management, preservation, and auditability.
A tool can be enterprise-approved and still not be legal-ready. Members noted that legal departments often need a higher defensibility standard than other business functions, particularly where prompts, outputs, transcripts, records, or agent actions may later need to be explained, preserved, or collected.
5. Build-versus-buy is becoming an orchestration question
The build-versus-buy conversation has become more nuanced. Members described buying legal data or point solutions where specialization matters, building internal tools where context and flexibility matter, and using orchestration layers to make the experience coherent for users.
One member framed legal data as a modular component that can be plugged into internal AI systems. Others focused on a single legal front door or orchestration layer.
6. Vibe-coded apps create opportunity and risk
Members were excited by the ability for lawyers to prototype or build lightweight apps, but they also saw a serious governance challenge. Attorney-built apps can help teams understand workflows and move faster, but they can also create unsupported tools connected to sensitive datasets, with unclear validation, access controls, ownership, and interpretation standards.
The key issue is not whether lawyers should experiment. It is how to give them room to experiment without creating uncontrolled legal technology sprawl.
7. Measurement needs to mature beyond usage
Members agreed that adoption is not enough. Legal teams need to know what tools are being used for, whether they improve cycle time or quality, whether they reduce outside counsel cost or manual work, and whether they reduce or increase risk.
Several members emphasized defining metrics up front. Without a baseline, teams can make almost any post-rollout story sound successful. Baseline capture, before-and-after comparison, and clearly defined success criteria were repeatedly treated as essential.
That emphasis on measurement reflects a broader pattern in Factor’s 2026 GenAI in Legal Benchmarking Report: directionally, the leadership network cohort, including members of The Sense Collective and The Vanguard Network, was far more likely to prioritize metrics and ROI tracking than the overall benchmark sample — 70% vs. 34%.
*Factor's GenAI in Legal Benchmarking Report.
The Sense Collective is Factor’s curated community for legal and innovation leaders advancing AI in legal.
