The conversation about AI in legal practice has shifted from whether to how. Law firms across Europe are now making real decisions about which tools to adopt, how to integrate them into existing workflows, and how to manage the associated risks.
This article outlines the key considerations for managing partners approaching AI adoption strategically.
Billable time drain
Thomson Reuters estimates attorneys spend 40% of their time on administration — not practice. AI offers the first real exit ramp.
Start with a workflow audit, not a vendor pitch
The most common mistake firms make is evaluating AI tools before understanding which problems they actually need to solve. Before speaking to any vendor, map your three highest-volume, lowest-value tasks — typically:
- First-pass contract review and clause extraction
- Legal research and case law summarisation
- Document drafting from precedents
These are the areas where AI delivers the clearest ROI and the lowest risk. Start there.
Due diligence on AI vendors for law firms
Not all AI tools are equal, and the bar for legal practice is higher than for general business use. When evaluating a vendor, ask:
- Where is data processed? EU-based processing is essential for GDPR compliance. Avoid tools that route data through US servers without adequate safeguards.
- What is the hallucination rate? Ask for documented benchmarks on legal tasks, not general performance metrics.
- Is the model fine-tuned on legal data? General-purpose LLMs perform significantly worse on jurisdiction-specific legal language.
- What audit trails exist? Partners need to be able to explain AI-assisted work to clients and courts.
The EU AI Act implications for law firms
The EU AI Act, fully applicable from August 2026, creates specific obligations for law firms using AI tools:
- Law firms using AI for legal research fall under the general-purpose AI system user category and must maintain basic documentation.
- AI tools used in litigation support or regulatory advice may qualify as high-risk systems, triggering more stringent requirements.
- Firms acting as AI deployers (using third-party AI tools in client-facing work) must implement human oversight mechanisms.
This is not a compliance burden to defer. Firms that build governance frameworks now will have a structural advantage over those who scramble in 2026.
Change management is the hardest part
The technical integration of AI tools is almost always easier than the human one. Expect resistance from:
- Senior associates who built their expertise on the tasks AI now automates
- Partners who are sceptical of tools they don't understand
- Clients who have concerns about confidentiality and quality
The most successful implementations I have observed treat AI adoption as an organisational change project, not an IT project. That means dedicated training, clear governance policies, and visible leadership commitment.
1%
Law firm R&D spend
vs 4% cross-industry average
$190M
Harvey AI ARR
Up from $50M in a single year
800 hrs
Capacity recovered
2 hrs/week x 10 lawyers x 40 weeks
90-Day AI Adoption Checklist
0/0Why Local Firms Are More Exposed Than They Realise
The structural reasons law firms resist AI are not simply cultural. Surveys of more than 35,000 attorneys over 35 years (Dr. Larry Richard, Caliper Institute) consistently show that lawyers score unusually high on scepticism and disagreeableness, and unusually low on resilience and openness to new ideas — the inverse of an innovator profile. This is not a character flaw; it is a feature of professional selection and training. Understanding it is the starting point for designing adoption programmes that actually work.
The financial structure compounds the problem. Law firms spend approximately 1% of operating costs on R&D versus a 4% cross-industry average. Under cash-basis accounting, every dollar of R&D investment is a direct hit to current-year partner distributions — there is no amortisation. The structural default is inaction, and healthy current margins make urgency invisible. The warning that should land with managing partners: "Do not become a forced buyer of technology. The time to invest is while margins are healthy."
The competitive threat is becoming specific. AI-native firms — startups building legal services from an AI-first foundation, such as Crosby (Sequoia-backed) in the US and Garfield in the UK — are valued at technology multiples, not law firm multiples, and can now compete with mid-market quality at a fraction of the cost. Lev Loukhton, a former Linklaters M&A partner now investing in legal tech, offers a concrete timeline: 2025 — no material revenue impact; 2026 — very modest; 2027 — non-trivial and real. That is eighteen months away, not a decade.
Is Your Firm Ready for AI?
0 questions
A phased adoption model
Rather than a firm-wide rollout, I recommend a three-phase approach:
Phase 1 — Pilot (3 months): One practice group, two or three volunteers, clearly defined use cases. Measure time saved and error rates.
Phase 2 — Guided expansion (6 months): Roll out to additional groups based on pilot learnings. Appoint a Legal Technology Lead or AI Champion.
Phase 3 — Governance and optimisation: Establish firm-wide AI use policy, vendor review cadence, and ongoing training programme.
Start With Work That Cannot Be Billed
The most common partner objection to AI adoption is that it cannibalises revenue: if AI reduces a four-hour task to fifteen minutes, who captures the difference? The answer, for the first phase of adoption, is simple: start with tasks that are not billed to clients at all.
Administrative and non-billable work — client intake interviews, matter intake forms, billing narrative enrichment, internal compliance checklists, email drafting — consumes a substantial share of fee earner time and generates zero revenue. AI deployed here faces no revenue cannibalisation objection. The gains are pure capacity recovery.
This reframes the conversation from "will we lose revenue?" to "will we free up time for client work?" In a 10-lawyer firm, recovering two hours per fee earner per week through non-billable automation adds back roughly 800 hours of capacity per year — before touching a single billable task. That is the pilot to run first.
Once the firm has absorbed how AI tools work in a no-stakes environment, extending to billable tasks — research, first-draft contracts, document review — follows naturally, with staff who have already overcome the learning curve.
The 90-Day Transformation: A Concrete Roadmap
For firms ready to move beyond the three-phase model, the 90-day sprint offers a more granular path:
Days 1–30 — Admin automation. Intake forms, email triage, secure Enterprise AI accounts. Goal: save 5–10 hours per week on pure administration. This addresses the "40% billable drain" — the proportion of attorney time consumed by administration, not practice. AI offers the first real exit ramp.
Days 31–60 — Research and drafting. AI-assisted research, first-draft generation, clause library development. Goal: reduce drafting time by 50% on standard document types.
Days 61–90 — Marketing and analytics. Client development content, financial analytics, matter profitability tracking. Goal: generate measurable business development output from AI tools.
The key insight from firms that have completed this sequence: start with a personal-interest hook. If a tool auto-generates perfect timesheets, you get 150% adoption because it directly serves every lawyer's self-interest. Ogletree Deakins, a 900-lawyer US firm, grew their innovation function from under $1M to over $16M in revenue by following this principle — and their most successful adoption strategy was "lawyers teaching lawyers" rather than IT-led training.
What Your Corporate Clients Are Already Demanding
The client side of this equation is moving faster than most firms realise. Mark Smolik, CLO at DHL and a Fortune 500 benchmark for in-house legal expectations, reports that clients are now asking a specific question in invoice reviews: "How much did you save working on my matters through generative AI last quarter?" This is not a general conversation about innovation. It is a line-item query.
DHL has gone further — building generative AI tools in-house to automate EEOC responsive pleadings end-to-end, pulling HR data and drafting first responses without outside counsel involvement. In-house departments are no longer waiting for firms to innovate.
Harvey AI's ARR grew from $50M to $190M in a single year, with a rumoured $11B valuation. That velocity tells you the demand side is real and accelerating. The question for a Basel firm is not whether AI-assisted legal services will be expected — it is whether you will be the firm delivering them, or the firm being replaced by one that does.
Adriana Adafinoaiei advises law firms on AI adoption, legal technology strategy, and EU AI Act compliance. To discuss your firm's situation, get in touch.