BMW builds some of the most sophisticated machines in the world. Its legal department is now investing in the technology behind its own operations.
Five years ago, those two sentences would have read as a mismatch. Today, barely.
On 8 October, Legora announced that BMW Group is rolling out its agentic legal platform across the carmaker’s in-house legal department. BMW’s Legal, IP and Compliance teams will use the platform to accelerate research, review contracts and litigation, monitor regulatory developments, and generate first drafts against their own precedents and standards. The deployment runs inside the tools that the team uses, including Microsoft Word, and fits within BMW’s access, security and governance frameworks. It’s being supported by a structured implementation programme run with Legora’s legal engineers.
The announcement notes BMW Group’s rationale behind the decision. In the press release, BMW’s Group General Counsel Dr Andreas Liepe explicitly linked AI adoption to “expanding our capacity to handle work in-house and maintaining our clear focus on the most complex and strategic matters.” This is a telling public statement from the person responsible for sending the department’s work either in-house or out, in effect saying where the work is planned to take place over the next few years.
Those sentences will no doubt be noticed by every DACH law firm partner with BMW on their client sheet.
The Legal Wire spoke with Björn Franke, Regional VP for DACH at Legora, four days after the BMW announcement and shortly after Legora opened its Munich office on 29 September. Björn joined Legora in June, having previously spent his career at BCG, where he led large-scale transformation projects across TMT, PE and automotive, including diligence work on software and growth assets.
We discussed BMW’s rollout, how corporate legal departments are being redesigned around AI, where that redistributes work between in-house teams and external counsel, and what Björn thinks the in-house function looks like three to five years from here.
The corporate legal department is changing
What does a modern GC do all day? Business strategy, operational efficiency, technology integration, risk management, often all at once, and often on the same matter. The job description has crept outward over the past decade without a formal reorganisation to mark the change. BMW’s own framing in the press release reflects it: the function describes itself as a strategic partner to the business, combining legal expertise with technical and commercial context, holding a mandate to identify and manage legal risk early. The team has been shortlisted for the azur Award for Innovation in 2026 and nominated as in-house legal team of the year at the JUVE Awards 2026. The GC’s job, in short, now looks a lot like a COO’s with an added licence to practise.
Björn’s background is a useful lens on that evolution. He spent his BCG years embedded in strategy work, where the discipline is to come into a company with an outside-in view, read the strategic context, and build solutions on top of that reading. He describes bringing the same muscle to Legora. “If you think about an in-house legal department, how does that tie into the strategy of that legal department? How does that tie into the overall strategy of the company? Maybe the AI strategy of the company? Who are the stakeholders who are driving that change?” Technology adoption inside a sophisticated corporate legal function, as he explains it, is a change-management problem, at least, before the issue of product is raised.
The muscle transferred.
The impact, in his view, is that the decision for or against any platform now hinges more on whether the client believes the vendor will be present through the whole deployment. Previously, it related more simply to the platform’s feature set. “The last thing is: do those clients or partners believe that we are not there only in the pilot, but are we really sitting next to them at their institution and putting those use cases into the platform? For that, you need to understand that change management and that people aspect are highly important to bring that change.”
TLW: In the legal AI deployments you’ve seen so far, what’s the most common reason a rollout stalls or underdelivers after a good pilot?
Björn “Rollouts can lose momentum when users return to their normal workload and no longer have both the support provided in the pilot as well as the time to work it out. Continued support should focus on real matters: showing where Legora fits and adapting existing processes.
At Legora, legal engineers help teams turn successful experiments into repeatable workflows and revisit use cases that have stalled. The aim is for legal professionals to be using legal AI confidently and independently in everyday work, and sustained adoption requires consistent engagement beyond initial training.”
What happens to the work?
So, where does the work go? Expanding internal capacity, by definition, means some matters that once went out to external counsel will now be handled by the in-house team. Whether that is 5% of a GC’s budget or 30% is the question incumbents will no doubt be trying to guage, and the answer will of course differ across companies, matters or jurisdictions.
Björn’s view, when asked whether this creates tension between Legora’s law firm clients and its corporate clients, is that both constituencies have more work than they can handle. “Generally there’s an increase globally in the volume of legal matters. I think that’s an underlying trend. And I think the second thing is that there’s a shift of higher-complexity tasks being captured by law firms in the future.” His framing is a two-tier distribution: standardised, repeatable work moves in-house with the help of AI, while higher-complexity, higher-stakes work continues to flow outward to external counsel. The overall pie, he says, is growing fast enough that both sides keep eating.
A reassuring reading, althought it also has an edge. Under this thesis, the implicit test for a law firm is whether its work is above the line of what an AI-enabled in-house team can handle on its own. Firms whose book of business is weighted toward standardised corporate workflows may be the first to feel the shift. Firms concentrated in genuinely complex matters may find that their in-house counterparts arrive better-prepared, with sharper briefs and tighter questions. That compresses scope, which is a kinder fate than killing scope, but a fate all the same.
TLW: If BMW’s rollout follows the pattern you’ve seen with other DACH corporates, which categories of work tend to move in-house first and which stay with external counsel even after Legora is embedded?
Björn: “Often, contract review against established playbooks, first drafts, document analysis and initial research move in-house. This is where in-house legal professionals have expertise but lack capacity. External counsel remains valuable for specialist expertise, unfamiliar jurisdictions, independent advice or additional capacity, including on major disputes, transactions and investigations.
With AI, in-house teams can do more preparation on those matters, giving external advisers more focused briefs. Complex and strategic work remains central to the in-house team. The allocation depends on the organisation’s capabilities and the matter.”
The question for in-house teams is one of composition. BMW is not, as far as the announcement goes, planning to shrink its legal department. It is planning to let the same team handle more work, more consistently, in-house. That outcome scales differently from the pure headcount-reduction story that often gets attached to AI adoption in other professional functions.
TLW: If composition changes rather than headcount, what do you think the new in-house team will look like?
Björn: “A team would have more responsibility for designing and managing legal workflows: intake, tasks for the technology, lawyer review points and quality checks.It will also be more important for legal operations and colleagues connecting legal, IT and the business, combining legal expertise with an understanding of processes and technology.A team would have more responsibility for designing and managing legal workflows: intake, tasks for the technology, lawyer review points and quality checks.It will also be more important for legal operations and colleagues connecting legal, IT and the business, combining legal expertise with an understanding of processes and technology.Björn: “A team would have more responsibility for designing and managing legal workflows: intake, tasks for the technology, lawyer review points and quality checks.It will also be more important for legal operations and colleagues connecting legal, IT and the business, combining legal expertise with an understanding of processes and technology.
Lawyers will spend more time directing work and reviewing results. These capabilities may sit in dedicated roles or develop within existing positions, depending on the work the department needs to deliver.”
The tech behind the transformation
Legora’s architecture for the corporate use case has a few specific pieces, and Björn describes them in the sequence a real deployment runs through. Intake is first: the moment business units raise a legal question or hand over a contract to the department. “We’re investing so much in legal intake, to make these processes more seamless from the business to the legal department and back.” The platform integrates with iManage so that contracts flow in automatically, get worked on inside Legora, and are saved back without the lawyer having to shuttle documents between systems.
The agentic layer, which Legora launched in June, is on top of that infrastructure. Björn describes the step change as one of complexity. “We’ve made a huge step with the agent going live in June. It’s a step change in the complexity of work which can be done by Legora. The focus of the user is shifting to more high-value tasks. The agent really needs to do that groundwork, and the user, the lawyer, needs to then focus on reviewing what the agent has built.” Human-in-the-loop is non-negotiable in his current reading. “Ultimately, the user, lawyer, or you could say the golf professional, needs to take the shot. He or she needs to be accountable for what they deliver.”
As an avid golfer, Björn identifies a caddy-and-golfer analogy: the caddy reads the course, picks the club, warns you about the wind. The golfer still has to swing, and if the ball ends up in the pond, nobody is suing the caddy. Björn concedes the analogy may evolve, but it holds for now. “It will be interesting to see in the future to what extent maybe lower-risk tasks will be handled by an agent automatically. But I think that’s a bit more of a future picture.”
For the corporate legal team, this means the agent does not replace a step, but it does compress several of them. Research, review, drafting and reporting can be run as a sequence rather than as discrete manual jobs, with the lawyer reviewing the output rather than assembling the inputs.
TLW: What is the biggest technical obstacle preventing AI agents from handling more complex legal work autonomously?
Björn: “The main technical obstacle is reliably completing an entire legal assignment. Correct citations do not rule out a missed exception that changes the conclusion. Legora grounds the agent’s work in the matter’s documents and the team’s playbooks and standards.
Complete assignments are tested against criteria written by legal professionals to identify weaknesses and improve the system. Greater autonomy depends on clearly defined workflows, with review points for decisions requiring a lawyer’s judgment.”
The DACH context around this rollout is crucial. More than 150 law firms and in-house legal teams in the region now use Legora. German lawyers on the platform have gone from around 3,000 at the start of the year to more than 13,000. Munich opened on 29 September. BMW’s announcement landed nine days later. Call that planning, luck, or perhaps the Munich office earning its keep from day one.
The future corporate legal department
Björn’s forward view centres on reorganising their work. The direction of travel he describes is one in which standardised, repeatable legal workflows become invisible, running automatically through the stack from business intake through to filed document, while the human attention of the legal team migrates toward the matters that need it. That is a plausible indicator of what the in-house function of BMW might be like a few years from now.
He has a realistic view of what that transition demands. The infrastructure is a precondition, not a solution. The deployment work, the stakeholder mapping, the integration with the systems a legal team uses, and the trust that gets built across the first six to twelve months of a rollout are what decide whether the technology changes the work or sits unused in the corner.
TLW: Five years from now, what do you think will be the biggest measure of success for an AI-enabled legal department?
Björn: “The strongest measure of success in five years for an AI-enabled legal department will be down to how much of the business’ legal needs the department can meet to a high standard and in a timely manner to influence decisions. Assessing the speed of useful advice will also be important, as well as its quality and the total cost of delivering comparable work.
The in-house team will have to evidence how they use the capacity AI creates. For instance, are they addressing previously unanswered requests or becoming involved earlier in important projects?
Businesses will also be assessing whether material risks are identified while there is still time to respond. Closer integration into the business should make legal expertise available when it is needed.”
Legora’s expectation, in DACH at least, is that these first corporate rollouts are the template for how in-house legal functions in Europe reorganise themselves over the next phase of AI adoption. BMW is one of the more visible early signatories. Watching how the deployment matures, how much of the department’s work genuinely moves in-house, and what the external counsel relationship looks like on the other side of it, will tell us whether the thesis holds.
Precision engineering, in other words, applied to a part of the business most engineers never see.
