The AI Law Firms Making AI Work For Them
Tom Borman, LawFuel contributing editor
The truth is, if your firm isn’t at least playing with AI, you’re falling behind. This isn’t about robot lawyers replacing bar exams. It’s about smarter workflows, sharper insights, and letting tech handle the grunt work so humans can do what humans do best: think creatively and connect personally.
But that familiar mix of awe and unease is valid and understandable. Some firms have jumped in with both feet—and ended up with fishy contracts or accidental data dumps. Others have struck the perfect balance.
Generative AI (GenAI) is proving to be a valuable assistant for those legal firms with the foresight and ambition to make the most of it.
Familiarity with AI has grown to 80 percent among legal professionals, which is an increase from 74 percent in 2023 and lawyers are wanting to invest in AI tools for legal research, document creation and a multitude of tasks.
We took a look at 10 law firms, from Big Law to solo boutiques, to see what is worth copying and what you need to avoid.
1. A&O Shearman – Magic Circle, Magic Bots
A&O Shearman has piloted Harvey (OpenAI/GPT‑4 model) since early 2023, rolled it out to ~3,500 lawyers across 43 offices, processed ~40,000 queries, and now operates “ContractMatrix” and other AI tools via their Markets Innovation Group and Fuse hub. Joining with Harvey, the firm has developed multi-step reasoning AI agents for complex legal tasks.
Why It Works
They didn’t go rogue—there’s a structured innovation lab and governance. Plus, they acknowledge: AI hallucinates. Lawyers must validate everything .
Watchouts
Hallucinations and GDPR risks. A&O is handling that with strict non-client/pseudonymised datasets and human oversight .
Lesson
Build internal labs, integrate with real human teams, plan with risk in mind to wind up with the results you want.
2. Simmons & Simmons (UK) –“Percy” Builds Efficiency and Competitiveness
Simmons & Simmons have developed “Percy,” an internal AI powered platform – Percy AI – that is designed to improve both productivity and work quality, but also market competitiveness . It’s used by 87 percent of fee earners with ~4,000 daily prompts. They’ve even built a 100-person AI Group advising clients on AI governance.
Why It Works
It’s fully internal and mainstream among fee-earners—so AI isn’t a weird new thing, it’s a team member.
Watchouts
Building internal tools takes time and money, plus ensuring integration and trust.
Lesson
Internal-first AI can drive adoption. Make it widely accessible and useful for everyday tasks.
3. VWV (UK mid-tier) – Trainee-Supported AI Culture
Bristol‑based VWV, a 500-strong firm, launched an AI innovation program that centered trainees helping to leverage off the young lawyers by turbo-charging their productivity with AI. Result: 7 POCs, including tools halving time for note‑taking and report drafting; trained staff in prompt‑engineering; spent £250k
Why It Works
They didn’t just buy AI; they built culture, excitement, and permission to test. Trainees pitched and implemented—so it’s bottom-up. The firms trainees say they rate AI training so highly, they’d tell their peers to choose it over increased pay. Their AI skills are set to increase their value to law firms, providing a career-long salary boost.
Watchouts
They initially bought licenses too fast without cultural buy-in—and recovered from that mistake .
Lesson
Culture is critical. Engage your people before getting the licenses.
4. Garfield AI (UK boutique) – £2 Letters Get SRA Approval
Garfield AI launched as a pure-play AI law firm, receiving SRA approval this year and offering £2 debt chase letters and £50 claim forms—fully SRA-approved. A former litigator + quantum physicist setup, with judicial support and it has not only attracted high attention, but also good results.
Why It Works
Clear niche: high-volume, low-cost claims. Lean, cheap, fast.
Watchouts
These services need strict QC: every output is human-edited initially. Risk? Disrupting fee models and review pushback.
Lesson
AI can create entirely new business models—but regulatory oversight isn’t optional.
5. Orrick (US) – Digital Law Center & AI Law Tracker
Orrick launched the Orrick AI Law Center, tracking global AI regulation and providing free tools, letting clients see how AI can or will affect their business. They also run Tech Studio—a hub with 50+ form generators and legal education tools and resources to assist businesses.
Why It Works
They position themselves as thought leaders and innovators—not just AI adopters. Clients benefit without heavy cost.
Watchouts
Free tools need sustainable staffing and tech investment.
Lesson
Thought leadership can create long-term client loyalty and brand authority.
6. Paul Weiss (US Big Law) – Testing with Harvey
The firm began piloting Harvey in January 2023, onboarded later that year. The Law firm gave no hard metrics yet—time savings are still anecdotal; output review overhead is high, but Harvey is designed to design repeatable workflows with key safety features. However they also focus on the ‘holistic’ support on AI issues to clients with AI advice on risks and issues around the rapid advance of AI.
Why It Works
Selective deployment helps validate value. They’re early-adopter but cautious.
Watchouts
No clear ROI yet; review costs could erase gains but time will tell, as always.
Lesson
Pilot small, set expectations low. Measure everything—even “review costs.”
7. Cleary Gottlieb (US) – Acquired Springbok AI
In March 2025, Cleary acquired London-based Springbok AI, integrating it to boost legal operations. It is a rare move for law firrms to acquire tech companies or AI units like Springbok, but Cleary see it as a key move towards providing a unique edge to their practice. The firm also entered a strategic partnership with Legora, a legal AI company, to enhance collaboration and review of legal processes.
Why It Works
Buying proven tech + adopting it gives firm-wide capability starting with existing workflows.
Watchouts
Post-acquisition integration is never simple—data, adoption, and culture issues arise unless good processes and systems are in place.
Lesson
Purchasing a legal AI outfit is faster than building—but requires serious integration work to make it effective.
8. BakerHostetler (US) – First with IBM Watson “ROSS”
Back in 2016, BakerHostetler adopted ROSS Intelligence ‘digital lawyer’ in bankruptcy practice—the first to do so in the pre-AI age and the firm has continued to expand and advise upon AI issues, including advising 40 companies on AI issues centered on data collection and use, algorithm development, IP ownership and licensing, data privacy and security impacts. linkedin.com+5thetimes.co.uk+5theaccessgroup.com+5en.wikipedia.org.
Why It Works
They targeted a niche usp: bankruptcy law, where pattern recognition and research parried volumes.
Watchouts
Early-stage tools like ROSS had limited scope and support; some wound down later due to evolving tech and data quality issues.
Lesson
Early adoption is bold—but maintaining edge means reinvesting as capabilities evolve.
9. Shoosmiths (UK) – £1M Bonus for AI Usage
Shoosmiths has been another early AI adopter and went so far as to reinforce its position at the forefront of legal innovation by becoming the first major law firm to link a firmwide bonus to the use of artificial intelligence (AI) by its workforce. The firm has set a clear and ambitious annual target: one million Microsoft Copilot prompts to unlock a £1m bonus pot for staff in its new financial year.
Why It Works
It incentivises usage—not just investment—turning AI from “nice to have” into a performance measure.
Watchouts
Bonuses can drive misuse—like inflating prompts counts. Need careful calibration of metrics.
Lesson
Use incentives—but align them to outcomes, not just usage stats.
10. Small & Solo Firms (US & UK) – ChatGPTs & Clio Duos
Small players use generic AI (ChatGPT), Clio Duo, Paxton AI, Parley (visa apps), and virtual receptionists—boosting research, drafting, intake and other activities that can create massive performance and profitability lift. A game-changer for smaller firms as well as large, levelling the playing field forr the smaller firms.
Why It Works
Affordable, off-the-shelf tools get big impact for low spend. For example, Parley cuts visa prep time in half and doubled caseloads for immigration firms .
Watchouts
These tools aren’t specialized for complex legal contexts—hallucination risk, lack of jurisdictional awareness.
Lesson
Pick tools that match the bucket: if you’re small, buy the Lego, don’t 3D print your own.
Common Smart Moves (and Occasional Head-Scratchers)
| Strategy | What Works | Caveats |
|---|---|---|
| Pilot before rolling out | A&OShearman, Paul Weiss test Harvey in small teams | Risk of culture stagnation if pilot never scales |
| Internal ownership of AI tools | Simmons & Simmons, VWV deployed internal solutions | Costly and requires tech-team discipline |
| Bonus/incentive schemes | Shoosmiths drove usage through financial alignment | Danger: gaming the system |
| Boutique specialization | Garfield AI, Parley—target niche, high-volume sectors | Overshaves—tools can misfire in nuanced work |
| Acquisitions vs partnerships | Cleary bought Springbok; A&O collaborated with Harvey | Integration challenges abound |
- C
Final Take–Homes for Your Firm
- Start small, pilot wisely: Roll out AI in one practice area with clear metrics and choose the pain point you are wanting to address.
- Engage your people first: Bring them into design and testing—like VWV’s trainees. Firms like Paul Weiss took on AI but haven’t seen cuts in time spent (review overhead eats savings).
- Track outcomes, not buzzwords: Bonuses are fine—just make sure they map to real impact. Don’t chase shiny objects, but focus on what can make a difference early.
- Plan for review overhead: Human oversight isn’t just nice, it’s essential. Big firms trying to deploy across complex transactional workflows risk failure and wasted spend.
- Regulation matters: Pick tools with approvals in regulated spaces.
- Leverage off-the-shelf for solos: Generic AI and SaaS tools offer massive leverage and properly used can make a massive and early difference.
- Think long-term: Acquisition/built tools need integration, culture, training. The team needs to be onboard with what you are wanting to do and what you want to achieve.
- Boost thought leadership: Like Orrick, being public with free tools builds trust and pipeline.
- Remember that AI isn’t magic (although it can sometimes seem so) and you still need a driver and good human input and involvement.
Really diggin’ how firms are getting on board with AI, like Orrick with their digital law center. It’s a game changer for how legal services evolve. Props to them for leaning into tech.
Super curious about the VWV’s trainee-supported AI culture? How does it work in practice, and does it mean trainees get less hands-on experience, or is it the opposite? Anyone got insights?
While AI’s impact cannot be denied, I question if relying too heavily on technology like ROSS by BakerHostetler strips the profession of some critical thinking skills. Isn’t there a risk of losing the human touch in legal analysis?
See your point, but I think it’s more about balance. Combining AI with human oversight could enhance efficiency without losing the essence of law. It’s the future of the field, IMHO.
£2 letters getting SRA approval is something else. Next, we’ll have AI making coffee and doing the dishes at the office. Where do I sign up?
This is exactly why I’m excited about the future of law! Seeing firms like Cleary Gottlieb acquiring AI startups shows commitment to innovation. Super inspiring for someone just starting out. Kudos to LawFuel Editors for highlighting these moves.
Oh great, more AI in law. Because what the world needs is robots deciding if you’re guilty or not. Can’t wait for the first AI lawyer TV drama, ‘Better Call HAL 9000.’