Legal departments are facing a paradigm shift. While artificial intelligence in legal work has long been limited to assistance functions—summarizing texts, searching for clauses, proofreading—we are now witnessing the transition to Agentic Legal AI. These systems no longer merely support work, but actively operate and make decisions independently within defined guardrails.
This article is based on the webinar "Agentic Legal AI - The New Standard. dormakaba Shows Why" with Cedric Ruepp (General Counsel, dormakaba Group) and David Alain Bloch (Co-Founder, Legartis).
What is Agentic Legal AI?
Agentic Legal AI refers to AI systems that go beyond pure support functions and can make independent decisions within predefined frameworks. Unlike traditional legal tech solutions that merely provide information or perform simple analyses, agentic systems can:
- Independently conduct complex contract reviews
- Assess and prioritize risks
- Generate improvement suggestions and implement them directly
- Ensure consistency across thousands of documents
The crucial difference: Agentic Legal AI takes over operational processes entirely, while legal expertise can focus on strategic decisions.
Real-World Example: dormakaba Shows the Way
The Swiss dormakaba Group, a traditional industrial company with over 150 years of history, impressively demonstrates how Agentic Legal AI works in practice. The challenge was clearly defined:
The Starting Point
- 9 qualified lawyers (7.8 FTE) plus 3 Contract Managers
- Over 10,000 contracts per year
- Products and markets on all continents
- Diverse contract types: from simple sales contracts to complex SaaS solutions
The Problem
Die Rechtsabteilung kann unmöglich jeden Vertrag reviewen – das Business muss juristische Ersthilfe selbst übernehmen können.
The Critical Role of Contract Playbooks
The success of Agentic Legal AI depends on one fundamental element: the Contract Playbook. A playbook is the structured set of rules that defines a company's legal standards. It specifies:
- Which clauses are mandatory
- Which formulations are unacceptable
- Which risks are treated with what priority
- How to proceed in different legal jurisdictions
The Two Central Challenges with Playbooksie
1. The Articulation Gap
Experienced lawyers intuitively know what to look for in a contract. However, translating this implicit knowledge into structured AI instructions is a challenge. The question "What do you check in a SaaS contract?" is often answered with: "This, that, and that—but I know it when I see it."
2. Validation and Scaling
It's not enough to establish rules once. The critical question is: Has the AI truly understood the guidelines? Does the review work consistently not only for the first contract, but also for the hundredth and thousandth?
The Breakthrough: Automated Contract Playbook Creation
dormakaba reports a "quantum leap" through automated Contract Playbook creation. What previously required weeks to months of manual work—manually capturing clauses in Excel, team alignments, qualifying each requirement—now takes hours.
How the Agentic Approach Works
Phase 1: Knowledge Acquisitionissensakquisition
An AI-driven dialogue systematically captures requirements:
- Company context and industry
- Typical contracting parties and legal jurisdictions
- Critical business risks
- Specific requirements (e.g., data protection, liability, contract terms)
Phase 2: Automated Playbook Generation
Agents create for each review requirement:
- Precise definitions
- Categorization (mandatory/optional)
- Risk classification
- Automatic test sets for validation
Phase 3: Validation through F-Scores
The system tests itself: Using sample cases, it checks whether the AI correctly interprets the instructions. An F-Score of 0.95 means that in 95% of cases, the requirement is applied correctly.
Concrete Results from Practice
The implementation of Agentic Legal AI with outsourced contract review to the business at dormakaba shows measurable successes:
1. Time Savings in Contract Review
- A 100-page contract: Reduction from 1.5 hours to 30 minutes
- Focus on truly critical clauses instead of mindless reading
- Automatic identification and correction of deviations from corporate standards
2. Quality Assurance
- Consistent review across all contracts
- Elimination of oversight risks in routine reviews
- Traceability through complete documentation
3. Scalability
- Contract review can be globally outsourced to business units
- Junior colleagues can review complex contracts with playbook support
- Legal department focuses on strategic topics
Implementation: Learnings from Practice
Cedric Ruepp, General Counsel Europe, Africa/APAC at dormakaba, shares honest insights into the transformation process:
1. Trial and Error is Normal
"The temptation would be great to present oneself as a visionary. The reality is more trial and error. We identified pain points and tested various solutions."
2. Cultural Change Takes Time
"We've internalized centuries-old work techniques—interpreting texts, source research. Suddenly everything is supposed to work at the push of a button. I had to test myself first: Am I really that error-free? In contracts I had reviewed 4-5 years ago, I found points in each one that I would do differently today."
3. Engagement is Crucial
"Agentic Legal AI is like a powerful sports car—you have to learn how to handle it. It requires someone internally to drive it forward. Especially at the beginning, until the playbooks are created and people have switched to autopilot."
4. The Right Partner Makes the Difference
Important selection criteria:
- Flexibility in playbooks (different contract types, legal jurisdictions)
- Full availability in relevant languages (not just machine-translated)
- Dynamic product development
- Support by Success Manager with best practices
Strategic Perspective: 2025 and Beyond
The central question for legal departments is: How do we scale without linearly inflating costs and team size?
Agentic Legal AI is not a distant vision of the future, but a strategic topic at board level. The development progresses in clear phases:
Stage 1: Assistant AI (Past)
- Text analysis and summaries
- Clause search
- Proofreading
Stage 2: Agentic Legal AI (Present)
- Independent contract review
- Risk classification
- Automated correction suggestions
- Consistency monitoring
Stage 3: Autonomous Systems (Near Future)
- Automated contract negotiation
- Direct document adaptation
- Proactive risk management
Best Practices for Implementing Legal AI
Do's:
✅ Jump in the deep end – There is no "ideal time"; technology is constantly evolving
✅ Start with manageable use cases – Use NDAs or DPAs as learning fields
✅ Plan resources – Especially in the setup phase for playbook creation and change management
✅ Define success metrics – Measure time savings, quality improvement, scalability
Don'ts:
❌ Exaggerated savings promises – Be careful with presentations to CFOs with FTE reduction forecasts
❌ Buy tool and hope – Without active engagement, the potential fizzles out
❌ Theorize too long – Learnings come from practice
❌ Neglect quality standards – A bad playbook leads to bad results
Conclusion: The Shift is Inevitable
"Agentic Legal AI is coming—it's only a question of when we jump on this train," summarizes Cedric Ruepp. The quote from dormakaba's Chief Innovation Officer puts it succinctly: "If you don't use AI in your daily work, you're not as efficient as you could be."
The shift from assistant to agentic AI systems enables legal departments to:
- Automate operational mass
- Create capacity for strategic tasks
- Handle large volumes even with small teams
- Increase quality and consistency
Agentic Legal AI does not replace the lawyer. But it enables them to apply entrepreneurial judgment and contribute to complex strategic decisions.
Want to learn more about Agentic Legal AI? Start today with Legartis and experience for yourself how agentic AI systems transform your contract review.
