Legal became the first functional spoke connected to RWE's central Data & AI Hub, making AI4Legal the proving ground for the Group's federated AI model.

With Legora as its core platform, RWE brought its AI ambitions into a shared, governed environment, embedding AI into daily legal work across the Group.

With Legora, RWE:

• Replaced fragmented, tool-by-tool experimentation across operating companies with one platform, shared governance and a common set of standards.

• Established Legal as the first functional spoke off the central Data & AI Hub, creating a reusable blueprint for other functions.

• Built reusable workflows, playbooks and skills for recurring legal work – from automated document review in real estate transactions and AI-assisted clause review in regulatory compliance, to term sheet implementation into complex transaction agreements and structured clause benchmarking across contract portfolios to support negotiation strategy.

• Set Legal AI principles that keep accountability with the lawyer, giving the wider business a working model for responsible AI rather than a policy document.

The challenge: experimentation without collaboration

“Legal teams across our operating companies were experimenting with AI in fragmented, disconnected ways,” says Nick Möller, who leads the Business Value Stream in RWE’s Group-wide Data & AI Hub and is Program Lead for AI4Legal. “Individual initiatives, different tools, no shared standards. Knowledge, templates and document data were siloed, so we couldn’t scale learnings or leverage Group-wide data.”

For lawyers, this showed up as requests by email, checked for completeness, then worked through internally developed templates across different formats and tools, with the same information entered manually several times over.

“These processes were highly fragmented and often more time-consuming than the legal analysis itself,” says Jakub Nowak, a Legal Counsel in Group Digital Integrity focused on data protection. “That took hours, and the frustration was not only about efficiency. It was that the process left very little space to focus on what actually matters: the legal assessment.”

The objective, according to Nick, was structural: “one platform with shared governance and an AI-ready data foundation so that improvements made in any team benefit the whole organization.”

Testing before scaling

RWE runs a hub-and-spoke model for AI. A central Data & AI Hub acts as the group's center of excellence, setting strategic direction, governance standards and shared platforms, while business units scale solutions within their own context. The intent is Group-wide consistency without sacrificing local agility. “This gave us clear communication lines, scaling advantages, and a reusable blueprint for other functions,” Nick says.

Markus Frowein, who leads Group Digital Integrity with responsibility for data protection, AI governance and the legal function’s AI strategy, describes a cross-functional team supporting the journey “from the initial evaluation and proof of concept through governance review, vendor negotiations and the design of the Group-wide rollout.”

"Legal teams brought the use cases and professional context," Markus says. "The Data & AI Hub connected the initiative to RWE's broader AI strategy; transformation experts contributed expertise in data and analytics as well as change management initiatives; and our governance functions tested whether the platform could meet the standards we expected across the Group."

Before scaling, RWE ran several controlled proofs of concept against real legal use cases, involving Group companies and legal departments early. “The results were consistently positive, but we did not rely on enthusiasm alone,” Markus says. “We combined user feedback with security and AI-governance assessments, contractual safeguards and a structured rollout model.” Progress was gated on evidence rather than dates: “proven use cases, sponsorship, and structure in place, not just a calendar date,” as Nick puts it.

The last decision was the shape of the relationship. Wanting to stay in control of how the technology itself is designed, RWE agreed a close development partnership with Legora, with structured alignment and roadmap discussions built in. “In a market evolving as quickly as legal AI, long-term value does not come from locking everything into a static specification,” Markus says, “but from combining strong safeguards with the right mechanisms to evolve together.”

Adoption on the users’ terms

Jakub’s first impression of Legora formed at the initial presentation. “There was a Scandinavian cleanliness to the design, an intuitiveness that told you straight away that this had been built by people who actually understand how lawyers want to work. That’s rarer than you’d think.”

“The first challenge was adoption,” Nick says, “and we got there through a mix of creative, even playful engagement formats and a champions network to build peer-to-peer momentum.”

“We never led with the tool,” he says. “By beginning with real problems and designing solutions together with colleagues, people became co-owners. That’s the difference between genuine engagement and mere compliance: people commit to what they helped shape.”

The platform lowered the barrier further by meeting lawyers inside the software they already had open. For Jakub, Legora’s Word and Outlook add-ins are what he relies on most. “That might sound like a small convenience, but it’s actually quite fundamental,” he says. “I don’t have to context-switch, I don’t have to break my workflow, I don’t have to think about where to go. It’s simply there.”

The same adaptability runs the other way: workflows, skills and agents are built to RWE's own processes, and the roadmap partnership means the platform itself can be shaped around them.

The clearest proof of adoption came from an interruption: a gap between the end of the pilot and the full rollout. “The moment it truly became indispensable wasn’t a moment of using it, it was a moment of not having it,” Jakub says. “That was when many of us noticed most clearly how much the platform had already become part of our day-to-day work. The absence of the tool said more than any feature demonstration could have done. It’s a strange thing to say, but losing access to a tool - even temporarily - is probably the clearest possible signal that it has genuinely become part of how you work.”

From experiment to operating model

At team and department level, Jakub points to playbooks and workflows, particularly through Legora’s Skills functionality. “They allow us to standardize how we approach recurring tasks, which touches several things at once: governance, process optimization, and accountability,” he says. “When a process is properly documented and applied consistently, transparency improves almost automatically. And in data protection work, where the ability to demonstrate compliance is just as important as achieving it, that is particularly valuable.”

Asked whether the gain has been speed or quality, he says, “I would not separate the two. Both have improved. Legora has helped us work more efficiently without compromising legal rigour.”

“The most significant shift is probably in what I expect of myself and of others on the team,” he says. “Because Legora removes so much of the preparatory friction, I can go deeper into the actual legal analysis.” Groundwork on a new matter can now begin while another is still being finalized.

He also returns to a small illustration: the two-minute rule, the convention that any task taking under two minutes should be handled on the spot. “What Legora has done is quietly expand what fits inside those two minutes. I can address things today that would previously have sat in my inbox for days simply because the entry cost was too high.”

For Markus, watching from the leadership side, the aggregate change is that AI stopped being a project. “The most important change is that AI has moved from isolated experimentation into the day-to-day operating model of the legal function. Colleagues are using the technology daily and with growing confidence, and it is becoming embedded in legal workflows across the organization.”

Organization-wide value

Putting every legal team on one platform only pays off if what one team builds can be picked up by another. "A validated playbook, workflow, template or knowledge source can support teams across regions and business units repeatedly," Markus says. "This is where a common platform creates compounding value and can bring scalable usability and quality, especially in a diversified company and with globally integrated teams."

The raw material for all of this is RWE's own legal knowledge, and making it usable is a deliberate investment rather than a by-product. Once adoption was established, Nick explains: "The harder challenge became execution at scale: making our proprietary legal data and knowledge assets, such as precedents, playbooks, and institutional know-how, globally accessible and AI-ready. That requires dedicated delivery teams made up of experts from different areas to operationalize use cases against that data foundation."

Adoption, in other words, is not the finish line. "One of the clearest lessons is how closely adoption, change management, governance, and data quality are connected," Markus says. "Introducing AI is not simply a technology deployment, it’s an organizational change process that requires coordination across multiple functions."

The goal is a legal function where one team's validated output is the starting point for another's, with regional teams operating as a single legal community rather than parallel ones.

A focus on responsible AI

RWE has positioned Legal as a driver of responsible AI across the whole organization. “Legora can support legal work, but it does not replace legal accountability,” Markus says. “Lawyers remain fully responsible for workflows, judgments and outcomes, regardless of how advanced the technology becomes.”

That is codified in the company-wide aligned Legal AI principles for the function, backed by an approved enterprise environment, technical and organizational controls, training and clear guidance. “A broad range of use cases sits in scope”, Markus says, “including reusable workflows and agents.” What doesn’t sit in scope is “delegating final legal judgment to the system or treating AI-generated output as automatically correct.”

“Responsible innovation requires clear guardrails, continuous adaptation, and ongoing training and communication,” Markus says. “Our goal is to make responsible behavior a natural part of day-to-day legal work.” Beyond the function, Legal’s job is to translate regulatory requirements, ethical principles and risk considerations into a framework the business can act on, while keeping governance “practical, proportionate and usable in day-to-day business.”

Markus describes the approach as pairing technical and organizational controls with intensive training and clear communication at all levels, so people build confidence rather than just comply.

“Trust is strategic,” Markus says. “We operate critical infrastructure and work across multiple jurisdictions. A federated and integrated hub-and-spoke model allows us to maintain common principles, platforms and governance at Group level while enabling individual businesses to implement solutions in ways that reflect their own operational context.”

At desk level, Jakub describes it as ownership. “The person who puts their name to a document or legal assessment is accountable for it, regardless of the tools used to produce it. In a sense, we’re all becoming reviewers. Increasingly, work comes to us as a draft to be assessed rather than a blank page to be filled.” He compares it to working with a capable colleague whose track record shapes how closely you check their output, “but I apply a somewhat sharper degree of scrutiny. The stakes in legal work are high enough that skepticism remains a professional virtue.”

He raises one consequence that rarely gets discussed: decision fatigue. “The more you use a platform like Legora - particularly as the agentic capabilities develop - the volume of micro-decisions increases. Risk-based prioritization becomes essential. Not every decision deserves the same weight of attention, and developing that judgment is part of using AI responsibly in legal work.”

The result: AI at the heart of legal workflows and services

“Our high adoption and frequent user levels show that we have put AI at the heart of legal workflows and services,” Markus says. What sits in scope from here is deliberately broad, extending to reusable workflows and agents built on the group’s own legal knowledge. Nick describes the next phase as “moving beyond adoption toward value realization and rethinking Legal’s operating model around AI”, with Legal’s blueprint now available to every function that follows it.

For the teams already working this way, the standard itself has shifted. “Once a team has internalized a certain standard of work, you can’t ask them to go back,” Jakub says. “The baseline has moved, and that’s not reversible.”

He sees a second return that matters as much as the work itself: the chance for colleagues “to stay genuinely current - to witness, firsthand, how the legal market is evolving and what it means to work at its frontier. Using the platform is not a passive experience, it’s part of learning how to work effectively with AI and how to shape its use responsibly in a legal environment.”

Which is why his advice to legal leaders still deciding is clear: “We got on the right train at the right time. And the advice I’d give to anyone watching from the platform is simple: don’t wait for the next one.”