Innovation Engineering: How Firms Can Think Bigger and Ask Better Questions
By Mike Raposa. Originally published on LAW.COM
There is a particular kind of fatigue setting in across the legal industry, and you can feel it at every conference. Every firm is talking about transformation. Every vendor is talking about change management. Every keynote ends with the same exhortation: the future is here, you must adapt, the firms that move fastest will win. None of it is wrong. All of it has become noise.
The problem is not that the advice is bad. It is that it has no shape. Telling a managing partner her firm needs to embrace transformation is like telling someone they should be healthier — true, unhelpful, and impossible to act on. What is missing is not more urgency. It is a better question.
Efficiency has a ceiling
For two decades, the conversation about law firm operations has revolved around efficiency. Faster intake, less expensive document processing, tighter service levels in the mailroom, the copy center, hospitality, reception, records. These are real services, and firms are right to want them run well.
But improving how that work is delivered, valuable as it is, eventually runs into a ceiling. When every provider is shaving the same costs and the same minutes year after year, the gains shrink, and no firm pulls meaningfully ahead on operations alone.
One answer has been consolidation — bringing many of these functions together under a single, accountable provider and connecting them through shared workflows and technology. Done well, that is real progress: it strips out friction, raises consistency, and gives a firm one partner to hold responsible for work that has to run flawlessly. What consolidation does not do, on its own, is change the nature of the work itself. It delivers the same work better and more broadly — a genuine gain, and a different thing from reinventing how the work gets done.
That distinction — between delivering the existing work well and transforming the work itself — matters more than ever right now, because the same instinct is beginning to shape how firms approach AI.
Most firms are aiming their newest tools at the work they already do — pouring their most powerful technology into running the same tasks a little faster. But when everyone automates the same tasks at once, no one pulls ahead: the bar resets, and the savings get competed away. That reaches the future a little faster while leaving a firm’s largest opportunity untouched — and that opportunity isn’t doing more of the existing work, but transforming how the high-value work gets done.
Technology usage is not transformation
The spending is real and the adoption is real, which is exactly why the missing transformation should worry us. Law firm technology spending grew 9.7 percent in 2025, the fastest real growth the industry has likely ever recorded, according to Thomson Reuters and Georgetown Law. In Williams Lea’s 2025-2026 survey of C-suite executives and managing partners across AmLaw 200 and other large firms, conducted with Sandpiper Partners, roughly nine in ten leaders reported actively engaging with generative AI and three-quarters already had formal AI governance policies. The tools are bought, the policies are written, the pilots are running.
And yet the value is not showing up. Asked about AI’s likely economic impact on their own firms over the next year, only 37 percent of those leaders expected cost savings, while half expected AI to increase their costs. More firm leaders expect AI to cost them money than to save it — even as managing client expectations around AI ranks among the top pressures they face. That is not a technology failure. It is the signature of an industry experimenting with powerful tools and no plan for turning them into value.
What separates the few firms capturing real value? The evidence is unusually direct. In McKinsey’s 2025 State of AI research, the single strongest predictor of enterprise-level AI impact was not the model, the data, or the budget. It was whether the organization fundamentally redesigned its workflows when it deployed AI — and high performers were 2.8 times more likely to have done so.
Set that against our own survey, where firm leaders named change management and internal resistance as the number one barrier to AI adoption, ahead of cost and data quality. The two findings meet in the middle: the winners are the ones who redesigned how the work gets done, and the thing stopping everyone else is the organizational difficulty of doing exactly that. The differentiator was never the tool. It was the work of rebuilding the work around it.
The operational layer no one is building
To understand why workflow redesign is so decisive, and so rare, it helps to separate what is actually happening inside a firm into three layers. Most of the industry’s attention, money, and anxiety is concentrated on the first two. The third is where transformation actually lives, and it is the one almost no one is building.
The first is the AI frontier model. Every firm has now picked one, whether it is a version of GPT, Gemini, Claude, or something else. This is the raw intelligence — the engine that reads, drafts, summarizes, and reasons. It is also, increasingly, a commodity. The leading models are converging in capability, they are available to everyone on roughly the same terms, and no firm builds its own. Whatever advantage exists here is shared by the entire market the moment it appears. A firm cannot differentiate on the model any more than it can differentiate on having electricity.
The second is the surface layer. This is the application that wraps the model and puts it on the lawyer’s desktop — the legal-specific interface every vendor is now fighting to own. It is crowded, well funded, fast-moving, and genuinely useful. But it is also, for the firm, another thing to buy. The surface tool arrives as a license, and a license is a capability waiting to be activated, not a capability in itself. Owning the best interface on the market does not tell anyone in the firm what to do with it on a Tuesday morning.
The third is the operational layer, and it is the one that decides whether any of the investment above it produces anything at all. It is the connective system between a firm’s chosen tools and the actual work of the firm: the redesigned workflow that decides which steps a model handles, which a person still owns, and how the handoff works; the institutional knowledge of how this firm runs a matter — how a filing actually moves, where the bottlenecks really are, which exceptions break the process; the human expertise that knows which parts of the work are judgment and which are merely effort. This is where a tool stops being a license and becomes a capability. It is, precisely, the fundamental workflow redesign the research identifies as the single largest differentiator between the high performers and everyone else.
The difficulty is the whole point. The frontier model and the surface tool can be bought; they arrive the same way for every firm that writes the check. The operational layer cannot be bought, because it does not exist in general — only in the specific. There is no off-the-shelf version of how this firm handles a contested matter intake, or how this practice group moves work between partners, associates, and support. It has to be built by hand, inside a particular firm’s operations, by people who understand both the technology and the way legal work really moves. That is slow, unglamorous work, which is exactly why the market has rushed past it to fight over the surface instead. The surface layer is where the demos happen. The operational layer is where the results do.
It also explains the most common failure in the industry right now. A firm buys a powerful tool, runs an impressive pilot, and sees no meaningful change six months later. The model works and the tool works, but no one built the operational layer connecting them to how the firm actually operates, so the tool sits on the desktop — capable, expensive, and idle. They are standing in front of a wall of options that all sound good, and so they reach for nothing. The investment was real. The transformation never had a layer to happen in.
What we mean by innovation engineering
Innovation engineering is the discipline of building that operational layer. It is not a product and not a frontier model. It is the work of sitting down with a firm, mapping how the work actually moves, and engineering the AI-enabled workflows that fit the way that firm operates rather than the way a slide deck imagines it should.
This is deliberate, hands-on, and consultative. It means a real conversation with a client’s technology and operations leadership about specific workflows, not a generic conversation about transformation. It means having teams with the expertise to sit across the table from a chief technology officer and design those workflows — something most providers in this market cannot do. The advantage is not a better model. It is knowing a client’s operations well enough, and having done this work long enough, to take whatever tools a firm has chosen, combine them with the human expertise around them, and turn that into something that genuinely changes how the work gets done.
Change is engineered, not installed
None of this is purely a technology exercise, which is the second reason it is hard. Real transformation is cross-functional by nature, requiring technology, change-management, and operational expertise brought to bear on the same problem at once; treating it as an IT project, or a leadership culture project, is one of the surest ways to ensure it goes nowhere. And resistance is not the enemy. Lawyers are not change-averse by nature; they are averse to change they had no part in shaping. That our own data names internal resistance as the number one barrier is not a sign lawyers are obstinate. It is a sign the industry keeps trying to install change rather than engineer it with the people who have to live inside it. The work of innovation engineering is as much about bringing those people into the design as it is about the design itself.
Be bold, think bigger, ask better questions
So here is the prescription, in the spirit of being more useful than the noise. Stop asking how to make your existing operations incrementally better. Start asking what your firm would look like if its core workflows were redesigned around what AI can now do — the one move the evidence most clearly rewards. Bring your operations, finance, technology, and risk leaders into the same room, because the answer lives across all of them and none alone. And demand a partner who will engineer the operational layer with you rather than sell you another tool for the pile. The firms that win the next decade will not be the ones that optimized the old model. They will be the ones bold enough to ask a bigger question.
The size of what is possible has changed. The firms that recognize that, and act on it with rigor rather than slogans, will define what comes next. We intend to be the partner that helps them do it.
