Legal teams need problem-first AI, not tech-first tooling

Legal departments face mounting pressure to adopt artificial intelligence, yet a significant gap exists between adoption and strategy. According to recent industry surveys cited in the article, nearly 87% of General Counsel report that their teams are actively using AI, but barely half have a formal roadmap for safe and strategic rollout. This gap highlights a common mistake: buying tools first and searching for use cases later. Kimberly Harlowe, Senior Director of Litigation Support and Technology at Altria, broke down a practical framework in a recent Data Xposure podcast episode to address this problem.

Harlowe's core principle is shifting from a tool-first to a problem-first mindset. When a colleague proposes a new legal technology solution, her first question is always: "What's the problem you're trying to solve?" Executive leadership—whether General Counsel or C-suite—cares about business outcomes: reducing risk, cutting costs, or freeing up internal capacity for strategic work. Pitching software features misses that audience. Instead, legal leaders should define the metric upfront (hard-dollar savings or reclaimed high-value capacity), assess the organization's risk tolerance early, and anchor proposals accordingly. A risk-averse leadership may need to see low-risk internal knowledge management tools succeed before approving litigation-facing automation.

The second pillar of the framework is designing for agility rather than locking in a rigid multi-year technology stack. Since AI solutions shift every six months, a department that commits to a single-vendor ecosystem risks missing superior tools down the line. Instead, Harlowe recommends focusing on the workflow problem (for example, "We struggle to keep up with subpoenas") rather than specific software brands. Solutions should integrate easily with existing legal operations ecosystems to allow tool swaps without breaking core processes. Critically, legal departments must cultivate a safe space to fail—teams need room to experiment, realize a tool isn't working, learn, and iterate quickly. Historically, legal culture penalizes mistakes, but AI testing requires a different mindset.

Building internal momentum with low-risk wins is the third step. A prime example is organizing legacy institutional knowledge. Corporate legal departments sit on thousands of legal memos, outside counsel advisories, and historical filings. Using standard enterprise environments like Microsoft SharePoint with automated AI fields, an automated prompt can parse incoming memos, extract key metadata (firm name, date, subject matter), and generate a concise two-sentence summary. The result: a chaotic document repository becomes an easily searchable knowledge base, drastically cutting lookup time without introducing external data security risks. These wins build tech confidence and literacy across the team, paving the way for larger initiatives.

The final and most impactful step is scaling high-impact workflows beyond internal operations to outside counsel, where the bulk of legal budgets are spent. Harlowe emphasized: "Your outside counsel are using tools every day and they're really expensive and the hours they're spending are huge. So I'm looking at how I can identify solutions that I can mandate use across multiple firms, because that's where I'm going to get real big bang for the buck." High-impact outside counsel use cases include bespoke deposition summaries (agentic AI tools parsing hours of deposition transcripts against core litigation issues), and written discovery drafts (AI tools using historical, approved discovery responses to generate first-pass drafts for new matters). The human-in-the-loop standard is non-negotiable: AI is meant to eliminate dozens of manual front-end hours so legal experts can focus on high-level strategy and precision review, not replace associates or partners. Surveying top law firms to understand what tools they are testing, then aligning around standardized AI tools, ensures consistency, improves defensibility, and delivers measurable cost reduction back to the business.

The article frames a fundamental tension in how legal departments approach AI adoption: technology-first thinking versus problem-first strategy. The data point that 87% of General Counsel use AI but barely half have a formal roadmap reveals a gap between early adoption and strategic governance. Kimberly Harlowe's framework addresses this by reorienting legal leaders away from vendor lock-in and toward agility—specifically, designing modular infrastructure that allows tool substitution as the AI landscape shifts monthly. This distinction matters because rigid, multi-year technology plans become obsolete quickly in a market where solutions evolve constantly.

The roadmap also reflects realistic constraints on legal operations: departments are small relative to their budgets, and executive leadership cares about measurable business outcomes (cost reduction, risk mitigation, capacity reallocation) rather than software capabilities. By starting with internal, low-risk automation—such as document organization—teams build both confidence and a proof-of-concept that justifies larger outside counsel initiatives. The emphasis on outside counsel scaling is strategic: it is where legal budgets concentrate, and where standardized AI tools can deliver the largest ROI. Throughout, the framework maintains a human-in-the-loop expectation, positioning AI as a labor-efficiency tool, not a replacement for legal judgment.

No comments yet. Be the first to share your thoughts!