As law firms and in-house teams sift past the hype around AI, the conversation is becoming more practical. The dialogue centers around how legal teams should evaluate it in a way that aligns with workflow, governance, and the realities of client-sensitive operations.
In this interview, Workstorm’s in-house associate speaks with its chief operating officer about what legal leaders should be paying attention to as they evaluate agentic AI. The discussion focuses on the considerations that matter most to law firms and legal departments alike: where agentic AI can create tangible value, what firms need to be aware of from a security and privacy standpoint, and why workflow should remain at the center of any long-term AI strategy.
Q: Agentic AI is getting a great deal of attention in legal tech. From your perspective, what should law firms be focusing on as they evaluate it?
The first thing I would say is that firms should distinguish hype from utility. The market is moving quickly, and there is no shortage of claims about what AI will do for legal teams. But for law firms, the real question is much more practical: where can agentic AI improve the way specific work gets done?
To answer that question, they must look beyond generic demonstrations and focus instead on bespoke workflow. Firms should ask whether an AI capability can help move work forward inside real legal processes such as during intake, making routing and assignment decisions, handling approvals, coordinating handoffs, initiating follow-ups, improving status visibility, and coordinating across teams. If agents cannot operate effectively in a real legal environment, their value will be limited no matter how impressive the technology appears in isolation.
The second consideration is accountability. Legal work does not happen in a vacuum. Firms need to understand where AI is participating, what role it is playing, and where individuals remain responsible for decisions, escalations, and oversight. In my view, the most promising use of agentic AI in legal contexts is reducing the operational friction around those judgments.
Q: When you say firms should evaluate AI through the lens of workflow, what does that mean in practice?
It means asking how the technology fits into the actual operating model of the firm.
Most legal work is not a single action. It is a sequence of steps involving different people, deadlines, approvals, dependencies, and exceptions. Firms should be looking at whether agentic AI can support that complexity in a controlled way. Can agents help route a matter correctly? Can they identify when a task has stalled? Can they support coordination across multiple participants? Can they improve visibility into what needs attention?
Those questions matter because they are tied directly to service delivery and operational performance.
Law firms should also be realistic about where structure exists and where it does not. AI tends to perform best when it is introduced into processes with some degree of predictability. If a workflow is highly variable, heavily relationship-driven, or dependent on nuanced judgment at every step, firms should recognize the role of AI may be more limited.
Q: What are the biggest mistakes law firms can make when assessing agentic AI?
One mistake is evaluating it as if it were purely a feature decision. Agentic AI affects how work is coordinated and advanced, making it an operating model question as much as a technology question.
I would also say firms should be cautious about confusing speed with operational fit. A tool may generate fast outputs, but if the output does not align with how the firm manages matters, teams, and client service, it can create more fragmentation rather than less, and more slop for teams to clean up than if they had just done the work without technology.
Q: Security and privacy are obviously central issues for law firms. What should firms be aware of when they evaluate agentic AI in this area?
Law firms should recognize that agentic AI introduces new considerations because it may be interacting with workflow, matter information, operational data, and systems sitting close to client work.
Firms should understand what information the AI is able to access, how broadly access extends across matters or systems, and whether the model is interacting only with the data needed for a specific task or with a wider body of information. They should be clear on where information is being processed, how usage data is handled, and what visibility exists into AI activity across workflows.
Firms should also pay attention to how AI participation affects confidentiality expectations, data boundaries, and internal governance. Specifically, look at workflow information, which includes matter names, participants, timelines, handoffs, and internal activity patterns.
Another area of awareness is platform dependency. If AI capabilities are deeply embedded inside a single environment, firms should understand what that means for data concentration, flexibility, and long-term control.
Q: Are privacy and security considerations different when AI is embedded into workflow rather than used as a standalone tool?
Yes, they can be.
A standalone AI tool may be used in a relatively narrow, user-driven way. When AI becomes part of workflow, its role can expand. It may be involved in routing, notifications, task management, status progression, or coordination across matters and teams. That means the scope of information, the frequency of interaction, and the operational footprint can all become broader.
Firms must understand where AI is active, what functions it is influencing, and what categories of information it touches as part of normal workflow execution. The more embedded the AI becomes, the more important it is to understand its role not just as a point tool, but as part of the firm’s operating environment.
Q: Beyond security and privacy, what else should executive teams in law firms be weighing?
They should be weighing durability.
The AI market will continue to change quickly. Models will improve, vendors will reposition, and new capabilities will enter the market at a steady pace. Law firms should be careful about making workflow decisions locking them too tightly to one model, one provider, or one narrow approach.
The better question is whether the firm’s workflow foundation is flexible enough to evolve as AI evolves. If a better model becomes available next year, can the firm incorporate it without redesigning how matters are managed? If the firm wants to use different AI systems for different functions, can it do so without fragmenting operations?
For executive teams, flexibility matters because technology decisions in this area promote adaptation over time.
Q: How does Workstorm fit into the conversation?
Workstorm is designed around workflow, coordination, and matter execution, which is exactly where many of these questions converge for legal teams.
Our view is firms should not have to place workflow inside a single AI product to benefit from AI. Workflow needs to be the stable foundation that gives firms control over how matters are managed, how teams collaborate, and how oversight is maintained. AI can then be introduced into the environment in a supporting way.
Workstorm offers flexibility. The platform can support multiple AI systems, agents, and tools within the same workflow environment, allowing firms to streamline their AI strategy without losing control of their process. It also supports the operating realities of legal work: structured workflows, cross-functional coordination, visibility into matter progression, and clear human involvement where approvals, escalation, or judgment are required.
Workstorm helps firms keep legal workflow and AI adoption aligned. Our objective is to make legal service delivery more coordinated, more visible, and more operationally efficient.
Q: For firms that are still early in evaluating agentic AI, what is the right mindset to take into those discussions?
Firms should approach agentic AI with discipline and the same way they would evaluate any change that affects client service, execution, and risk. This approach includes determining where workflow can improve, where it requires boundaries and guardrails, what it means for visibility and accountability, and whether it strengthens or complicates the firm’s operating model.
The firms getting the most value from agentic AI will be the ones treating it as an operational capability, not just an innovation initiative. The long-term winners will be the firms adopting agentic AI in ways that improve how work flows while preserving the standards their clients expect.
