For the past two years, the legal industry has been answering a single question about AI: "How many hours will it save?" That question was useful — it got budgets approved, pilots launched, and skeptics into seats. But in 2026, a growing chorus of legal leaders, CFOs, and technology vendors are saying the same thing: hours saved was always the wrong finish line.
At a recent industry panel on AI's real-world return on investment, practitioners from a Fortune 10 legal department, a midsize law firm CFO's office, and a legal tech CEO converged on a striking consensus. The old efficiency calculus — input cost, subtract time saved, claim ROI — no longer captures what AI is actually doing to legal work. The returns are broader, deeper, and in some cases, entirely new categories of value that didn't exist eighteen months ago.
Here's what the conversation revealed — and what it means for firms and legal departments trying to build a credible business case for AI investment in 2026 and beyond.
LegalWeek 2026 — "Beyond the Balance Sheet: Rethinking ROI for AI in the Legal World" panel featuring Jeff Reihl (LexisNexis), Afua Dabanka (META), Madhav Srinivasan (FBT Gibbons), and Stephanie Curcio (NL Patent)
The Three-Bucket Framework: Quantity, Quality, and Quantum
Perhaps the most useful lens offered during the discussion came from Madhu Srinivasan, CFO of Forvis Mazars' legal advisory practice, who proposed sorting AI's payback into three distinct categories.
Bucket 1: Quantity
The familiar one. Measurable time saved: fewer hours drafting discovery responses, faster document review, quicker first cuts of contract analysis. This is where most firms have parked their ROI conversations — and for good reason. It translates directly into hours, rates, and dollars.
Bucket 2: Quality
Harder to pin down but no less real. Are your work products more thorough? Are fewer issues slipping through review? Is the data you pull from financial systems cleaner, more contextual, more actionable? Quality improvements show up indirectly: in fewer revisions, stronger client relationships, and better outcomes at trial or settlement.
Bucket 3: Quantum
Work that simply was not possible before. Natural language queries across massive financial databases. Predictive analytics on client risk or write-off exposure. Automated compliance monitoring that scales across thousands of matters simultaneously. These are not efficiency gains. They are new capabilities, and their economic value may dwarf the first two buckets over time — even if the payback horizon is longer.
"The fixation on hours saved is what we're going to laugh about in three years. We should be reimagining entire processes to create new revenue streams, not just shaving minutes off the old ones."
— Panel Consensus, LegalWeek 2026
For firms building a business case today, the three-bucket model offers a critical reframe. Not every AI investment needs to show immediate dollar-for-dollar payback. Some investments are strategic bets on capabilities that will compound over time.
AI Is Collapsing the Law Firm Pyramid
One of the panel's most provocative insights came from Stephanie Curcio, CEO of NLPatent, who shared a trend that surprised even her own team: senior attorneys are now among the heaviest users of AI tools on her platform.
This upends the traditional law firm pyramid. For decades, the economics of practice have depended on leverage — senior partners win the work, junior associates execute it, and the spread between billing rates generates profit. AI is compressing that model. When a partner can use an AI tool to handle what previously required delegating to a second-year associate, the value proposition to the client changes fundamentally.
"You get me, not my junior."
— Client-facing pitch from a senior practitioner leveraging AI tools
That's a powerful differentiator. It eliminates the broken telephone effect of multi-layer delegation, puts the most experienced legal mind directly on the problem, and signals to clients that they're paying for judgment, not volume.
The downstream implications are significant. If senior attorneys can handle more work directly, what happens to the associate development pipeline? Firms are divided: some are restricting AI access for junior lawyers so they learn the craft first; others hand them the tools on day one. Neither approach has proven itself yet. But the structural shift is already underway, and firms that ignore it risk finding their economic model quietly hollowed out.
A Four-Metric Measurement Framework
Panelists from the corporate legal side offered a practical measurement framework that their analytics teams are actively piloting. It consists of four dimensions, each capturing a different facet of AI value.
Cost Optimization
Direct savings — license consolidation, reduced outside counsel spend, lower per-unit cost of document production.
Operational Efficiency
The classic hours-saved metric, but contextualized within specific workflows rather than measured in aggregate.
Transformation Velocity
What percentage of the organization's processes have been augmented by AI? An adoption-as-progress metric that tells you how fast you're moving toward AI-native.
Strategic Value Creation
New capabilities, new revenue streams, and new competitive advantages that didn't exist before deployment. The quantum bucket in measurement form.
The framework is still being tested, and the panelists were candid about that. But its structure reflects an important shift in thinking: from measuring AI as a cost-reduction tool to measuring it as a capability multiplier.
Good Invisibility: When AI Disappears Into the Work
One panelist introduced a concept worth borrowing: "good invisibility." The idea is simple but revealing. True success with AI doesn't look like a flashy demo or a celebrated pilot program. It looks like nobody talking about AI at all — because it's so deeply embedded in daily workflows that it's just how work gets done.
Good invisibility means lawyers aren't switching to a separate tab to access an AI tool. They're not logging into a different platform. The intelligence is built into their existing environment — surfacing relevant precedent during drafting, flagging inconsistencies during review, pre-populating fields from financial data. When adoption reaches that point, the ROI conversation becomes moot, because the tool is indistinguishable from the practice itself.
Getting there requires more than good technology. It demands thoughtful integration, genuine workflow redesign, and — critically — change management that goes far beyond button-and-feature training sessions. The most effective adoption strategies, panelists noted, rely on internal champions sharing real use cases and peer-to-peer storytelling, not top-down mandates or vendor-led feature tours.
The Build-vs.-Buy Pendulum Is Swinging Again
An emerging undercurrent in the market is a quiet resurgence of the build-versus-buy debate. Large firms and corporations with internal development resources are increasingly dissatisfied with general-purpose AI platforms that don't support specialized workflows — IP prosecution, complex family law, regulatory compliance in niche industries.
Rather than forcing a general tool to do specialized work, some organizations are exploring hybrid architectures: a standardized AI platform at the core, with purpose-built modules and API-connected point solutions layered on top. These bespoke integrations — often built quietly and internally — are designed to handle the exact workflows that off-the-shelf products can't.
The panelists cautioned, however, that building carries its own risks. Internal tools require ongoing maintenance, dedicated staffing, and the ability to keep pace with a technology landscape that is evolving at an extraordinary rate. The hybrid model attempts to balance customization with sustainability, but it's early days, and the long-term economics of that approach remain unproven.
For smaller firms and solo practitioners, the build option is rarely viable. What matters is choosing vendors whose platforms are designed for extensibility — tools that can adapt to specific practice areas, integrate with existing case management systems, and evolve alongside the technology without requiring a full rip-and-replace every eighteen months.
The Mandate Question
Should firms require lawyers to use AI tools? The panel was split — not on the value of adoption, but on the mechanics of achieving it.
Corporate Legal Departments
Mandates are more feasible. There's a clear hierarchy, and leadership can direct tool usage. Some departments have set explicit adoption targets — daily usage rates above 90% — and are actively monitoring license utilization to reallocate expensive seats from non-users to willing adopters.
Law Firms
The dynamics are different. Lawyers resist being told what to use. A bottoms-up approach — embedding AI into the tools lawyers already rely on, like time capture, research databases, and billing systems — tends to generate adoption more naturally than top-down edicts.
The deeper point is that adoption without engagement is a vanity metric. A lawyer who opens an AI tool every day but never moves beyond basic prompts isn't generating ROI. The firms seeing real returns are the ones investing in contextual training, internal communities of practice, and leadership that models AI-forward behavior — not just the ones counting logins.
What This Means for Legal Technology Strategy in 2026
The panel's collective message was clear: the legal industry's AI conversation is maturing. The experimental phase is over. The "hours saved" justification is necessary but insufficient. And the firms that will lead in the next three years are the ones that can articulate AI's value across all three of Srinivasan's buckets — quantity, quality, and quantum — while building the measurement infrastructure to prove it.
At LegalTek.ai, we see this evolution every day. The firms we work with are moving beyond point-solution thinking toward integrated, practice-specific AI architectures that don't just make existing work faster — they make fundamentally better work possible. That's the ROI worth measuring.
Three Predictions for Where This Is Heading
Task-level ROI will give way to practice-level transformation metrics
Firms will stop measuring AI tool by tool and start measuring how entire service lines have changed.
Client-facing differentiation will overtake internal efficiency
The winning pitch won't be "we're cheaper." It will be "we can do things no one else can."
Hybrid architectures will become the norm
General platforms plus specialized integrations, connected by APIs and orchestrated by agents, will replace the monolithic legal tech stack.
