The legal industry is drowning in AI hype. Every vendor promises transformation. Every firm is investing. But the data tells a different story: AI is saving time and money for some law firms while burning cash for others. The difference isn't the technology. It's whether firms actually know how to use it.
The numbers look good on paper. Lexis+ AI delivered 344% ROI for large law firms over three years according to a Forrester study, with the platform paying for itself in under six months. Corporate legal departments saw 284% ROI. By 2026, 78% of legal professionals are using AI, up from 27% in 2024. The adoption curve is steep.
But here's the catch: those Lexis+ numbers come from firms that already knew what they were doing. The broader market is messier. According to research from Thomson Reuters, law firms with a clear AI strategy are almost four times more likely to see tangible returns on investment. Flip that around — and it means most firms adopting AI without a strategy are probably wasting money.
The Real Time Savings (When They Happen)
The documented wins are specific. Lexis+ AI users reported 2.5 hours per week saved for senior associates and partners on drafting and research. Research staff saw 225 hours saved annually. Junior lawyers experienced up to 35% reduction in written-off billable hours.
Spellbook, which focuses on contract review and drafting, now has 4,000+ in-house teams and law firms using the platform. In its 2026 State of Contracts report, Spellbook found that AI adoption is reshaping contract terms themselves — 18% of SaaS agreements now include AI usage policies, a clause type that didn't exist two years ago.
But time savings don't automatically mean money in the bank. A lawyer who saves two hours per week on research still has to bill those hours to a client, or the firm loses revenue. That's the structural problem. As as we covered in AI contract review beats junior lawyers—so why aren't firms using it?, the billable hour model creates perverse incentives around AI adoption. Efficiency can hurt firm profitability in the short term.
E-Discovery: Where AI Actually Cuts Costs
E-discovery is different. It's not a billable service — it's a cost center. That means AI that reduces e-discovery costs directly improves margins.
The mechanics are straightforward: AI-powered platforms like Relativity, Logikcull, and Reveal Data use predictive coding and machine learning to identify relevant documents, flag privilege issues, and reduce the volume of data requiring human review. Early case assessment powered by AI can cut discovery costs by 50% or more according to platform claims, primarily through data reduction.
The time savings are real. AI can categorize, tag, and summarize thousands of documents in hours instead of weeks. It flags potentially privileged content before human review begins, reducing privilege log preparation time. Real-time evidence integration means lawyers don't waste cycles on manual data consolidation.
But here's the reality check: e-discovery platforms have been using machine learning for years. The recent wave of generative AI hasn't fundamentally changed the math — it's made the existing playbook faster and slightly cheaper. The bigger win is that more firms are actually using these tools now. Adoption of AI in e-discovery accelerated in 2025-2026 as the tools became less intimidating and more integrated into standard workflows.
The Adoption Problem: Why Most Firms Aren't Seeing Returns
Here's where it gets uncomfortable. According to the 2026 data, while 78% of legal professionals use some form of AI, most report only 6-20% weekly time savings. That's not transformational. That's marginal.
The problem: firms are treating AI as a tool to bolt onto existing workflows, not as a reason to rethink workflows. A lawyer using Lexis+ AI to speed up research but still billing by the hour isn't capturing the value. A firm using Harvey or Spellbook for contract review but not changing how it staffs deals isn't realizing the ROI.
Thomson Reuters warned in early 2026 that the AI bubble could burst for firms without clear strategy. The report found that while firms are spending aggressively on AI — increasing technology investments by 11% and knowledge management by 10% — the returns are uneven. Firms that implemented AI thoughtfully saw greater returns. Firms adopting AI mainly to justify higher billing rates or keep up with competitors are more exposed if enthusiasm cools.
The data is clear: law firms increased tech spending dramatically in 2025, but profitability gains were concentrated among firms with explicit AI strategies. Everyone else spent money and got marginal improvements.
Case Prediction: Still Overhyped
Case prediction AI — tools that forecast litigation outcomes — remains one of the most oversold applications in legal tech. The promise is obvious: predict your win probability before trial, use that to inform settlement strategy, reduce uncertainty.
The reality is murkier. Prediction tools work best on historical data from similar cases, which means they're most reliable in high-volume practice areas with consistent precedent. They're least reliable in novel cases or emerging areas of law — exactly where lawyers most want predictions.
The accuracy question is also unresolved. Most vendors don't publish independent validation studies. Some platforms claim high accuracy on their own test sets, but those claims don't always hold up when applied to different jurisdictions or case types. The ROSS Intelligence litigation against Thomson Reuters — which ROSS lost — highlighted the risks of overconfidence in AI legal research.
Prediction tools are useful for specific, narrow problems. They're not the game-changer vendors claim.
The Firms Actually Winning
The pattern is consistent across firms seeing real ROI:
Clear use cases. They identified specific, repetitive tasks where AI could measurably reduce time or cost. Contract review. Legal research. Document summarization. Not vague "AI transformation."
Workflow redesign. They didn't just add AI to existing processes. They rebuilt processes around AI capabilities. That meant changing staffing models, billing approaches, or client delivery models.
Measurement. They tracked specific metrics — hours saved, documents processed, cost per review, time to resolution — and adjusted tools based on results.
Governance. They built processes to ensure AI output quality, manage client expectations around AI use, and maintain compliance with evolving AI audit requirements.
Most firms are skipping steps 2-4. They buy the tool, train people on it, and hope for the best. That's why average time savings are so low despite the technology working well in controlled studies.
What Matters Now
The legal AI market is maturing past the hype phase. Vendors are consolidating. Firms are learning which tools actually work. The gap between leaders and laggards is widening.
If you're in-house counsel evaluating legal AI, don't ask "what tool should we buy." Ask "what process would change if this tool worked perfectly." If you can't answer that, the tool won't deliver ROI.
If you're a law firm, the competitive advantage isn't having AI. It's having a strategy for how AI changes your business model. Firms that figure that out first will capture the value. Everyone else will have expensive software and mediocre time savings.
Originally published on Derivinate News. Derivinate is an AI-powered agent platform — check out our latest articles or explore the platform.
Top comments (0)