The 30-second version: Everyone in legal is suddenly talking about legal engineers and forward deployed engineers, and almost nobody means the same thing. I've spent my career doing the jobs this role is stitched together from, process work, firm strategy, building things partners actually use, and this semester I'm turning law students into forward deployed legal engineers at IU Maurer. So this edition is a decode, not a hot take. I pulled the actual job specs from three corners of the market, a traditional firm, a neo firm, and the leading vendor, and they describe three different jobs. Read correctly, those conflicting specs are the best market intelligence you'll get all year, because all three companies designed the role from their own need outward. From there I name the need inside a firm, break the role into six components nobody has to max out alone, and explain why retitling people you haven't invested in is the worst move on the board. Ten minutes. You'll leave with signal.

THE FULL BRIEF

A title on its way to meaning nothing

You've seen the term by now. Legal engineer. Forward deployed engineer. FDE, if you spend time in some corners of X. The role came out of Palantir, got adopted by the legal AI vendors, and is now the subject of a small publishing industry.

We've run this play before in legal. A title gets hot faster than its definition is understood, and within two years it's a costume. Ask anyone who carried an "innovation" title through the last decade what the word was worth by the end of it. "Legal engineer" is next in line. Law.com says demand is skyrocketing. The ABA has an explainer. LinkedIn has a thousand people who discovered the term in March and became authorities on it by May, and every vendor with a content budget has published its own definitive guide to the role.

One reading note before we decode. Look at the bylines. The most polished piece in the genre, "The Rise of the Legal Engineer," was written by Legora's Head of Legal Engineering. Thomson Reuters is standing up a CoCounsel Forward Deployed Engineering team. Harvey, by one outside analysis this spring, has somewhere between 40 and 80 forward deployed engineers embedded inside firms like A&O Shearman, Cleary Gottlieb, and Reed Smith, on stretches of six to nine months. Nothing dishonest in any of that. But nearly every voice defining this role for you has a business model that needs the role living on their terms in their business.

Let’s dig in.

Three specs, three different jobs

Job postings are helpful. A posting is where a company commits, in writing and in public, to what it believes a role is, because the wrong words attract the wrong candidates and burn real money. Here are three, from three corners of the market. They describe three different jobs, and they can't even agree on what to call them.

Kirkland & Ellis, the traditional firm. Kirkland is hiring an Innovation AI Developer at $177,000 to $218,000. The spec asks for RAG pipelines, vector databases, LLM evaluation frameworks. Legal domain experience is listed as "preferred." This is more of an engineer's spec, with the law element optional.

Before you laugh, understand this. Kirkland isn't guessing. The firm is betting it can embed the legal half into technical hires, and it's running that bet alongside the company that invented the role: in June, Kirkland and Palantir announced an exclusive AI platform build for private equity fund formation, part of an AI push Bloomberg Law reports at half a billion dollars and hundreds of hires. Palantir didn't just originate the forward deployed engineer. It defined what the role does, wrapped in its favorite word, ontology, which sounds like philosophy and mostly means decoding your systems, your behaviors, and your management flows until the software fits the institution. Kirkland's need is to build, at scale, inside its own walls. So, it wrote a builder's job and bought the original playbook from the original author.

Now note what's missing. Nowhere in Kirkland's postings will you find "legal engineer" or "forward deployed." The titles are deliberately boring. Innovation AI Developer. AI Engineer II. The firm making the biggest disclosed AI bet in legal skipped the fashionable vocabulary entirely and procured the people who invented it, Palantir's actual forward deployed engineers, under contract instead. Hold that thought. It matters in a minute.

Norm AI, the neo firm. Norm, the quietly hot AI company backed by more than $100 million from Blackstone and others, wrote its spec in the opposite direction. Its "Legal Engineer, Forward Deployed" posting required a JD, two to four years of BigLaw or regulatory experience, and technical curiosity. Curiosity. Not engineering. Norm teaches the technology; what it's buying is legal judgment, a translator who sits between senior lawyers and AI engineers and decides what's worth building at all. Norm's need is translation, so it wrote a translator's spec. That posting, at $175,000 to $225,000 plus bonus and equity, is already closed. They filled the roles with mid-year associates out of Milbank and Skadden. And watch the direction the person traveled: out of a firm, into a vendor. The trade press spent the spring covering exactly this flow, AI legal tech startups pulling lawyers off the partner track. Your associate ranks are the farm system for the companies selling to you.

Harvey, the leading vendor. The third spec is the one to read deeper. Harvey's Legal Engineer role, per a posting circulating on LinkedIn this month, is for "former practicing lawyers who bridge customers, sales, and product to support all aspects of our sales strategy and product development." The requirements: a JD, at least three years in-house or advising corporate legal departments, executive presence, demo and workshop skills, and "curiosity about AI." That's the entire technical bar.

Then read the compensation line, because it confesses what the words dress up: $270,000 to $320,000 OTE on a 75/25 split. OTE means on target earnings. The split means a quarter of the money is variable, paid on performance. Nobody pays engineers that way. Sales teams get paid that way. Harvey's need is adoption and revenue, so it wrote a seller's job and gave it an engineering title. And catch the sourcing detail: this spec wants in-house experience. Harvey isn't just hiring out of your associate ranks anymore. It's hiring out of your clients' legal departments, to sell to your clients.

The signal in the noise

What we have? A builder, a translator, a seller.

Now watch what the title does across them, because this is the thought I asked you to hold. Kirkland, doing the deepest technical build, never uses "legal engineer" at all; its titles are plain to the point of boring. Norm uses the title with a qualifier bolted on. Harvey pins the shiniest, cleanest version of the title on the one job with no engineering in it, and pays that job the most, more than Kirkland pays the people building the actual systems. The vocabulary gets fancier as the work gets closer to the sale. That's bastardization measured in real time. It also tells you how much weight the title itself can bear, which is none.

If you stop there, the takeaway is "the term is meaningless" and you tune the whole subject out. That's the noise reading. The signal reading is better, and it's the point I am driving at. Nobody at Kirkland, Norm, or Harvey sat down and asked "what is a legal engineer and how do we get one." Each asked what it needed and designed the role from the need outward. Kirkland needs to build, so its version is a builder. Norm needs translation, so its version is a translator. Harvey needs adoption, so its version is a seller.

Here's how the bastardization will actually arrive: firms copying the title without thinking. Asking "how do we get a legal engineer." Posting specs assembled from other companies' postings. Or, cheapest of all, relabeling the KM director and calling the capability built. Designing from someone else's job posting inward is how you fail with AI adoption.

And I want to hit on that relabeling urge for a minute, because our profession has a title pathology that predates AI by decades. Law firms are notorious for handing out titles that don't reflect the actual job, or the agency behind it, or the authority. The law firm C-suite is littered with chiefs who carry real administrative and managerial burden and hold zero decision rights. Everyone in the building knows which chiefs can actually decide something. It's a short list. So, when the AI pressure arrives, the reflex will be to relabel a few good people, slap "legal engineer" or "AI" into a title, and hope the title changes the job.

It won't. A title doesn't add a single capability to the person wearing it. What it does instead is damage, in two directions at once. Outward: everyone else assigns the title meanings of their own, expectations the holder never agreed to, often doesn't fully understand, and can't deliver, and every gap between the label and the delivery gets scored against your whole AI effort. Inward: the title inflates. Now you've got someone out over their skis, acting with authority they don't hold, without the experience or skill to back it up. Partners can smell performed authority (faux expertise) from the far end of the hallway, and the first failed promise confirms every skeptic in the building. Both failure modes are cancerous to real adoption. And notice that neither one started as a talent problem. They started as sequencing problems.

The sequence that works is the unglamorous one. Invest in the people first. Give them a real job with a written scope and real work to rebuild, and be ruthlessly clear, to them and to everyone around them, about what they have the authority to decide and what they don't. The title can come later, once the capability is real and the building has watched it work. A title that arrives after evidence names something. A title that arrives before evidence promises something, and likely that promise will not be kept.

My advice is to ignore the question "what is a legal engineer." Ask the real one instead. What is the need inside your law firm or legal team?

The need, named

An MIT report last year put the failure rate of enterprise gen AI pilots at 95 percent, and the vendors' entire forward deployed model exists because they learned the lesson in that number: value dies in deployment, not in the software. The need most team have is a deployment capability. In practice that means sitting with the people who do the work and mapping how it actually happens, exceptions and all, because the process as documented is never the process as practiced. Building on the systems the firm already owns. Testing what gets built against evidence: pass rates, failure modes, cost per run, hours returned to fee earners. Bringing partners along so the thing gets used rather than admired. And keeping every lesson inside the walls, where it compounds, instead of watching it walk out when a vendor's engagement ends.

Design from that need outward and you get the firm side version of the role. Call it what it is, a forward deployed legal engineer: the same fused profile the vendors are fighting over, deployed in the opposite direction, into your own practices, on your own payroll, for your own benefit.

Two things before I break the role apart, because I don't want you reading what follows as something conjured by me in my own head.

First, everything old is new again. Most of the skills on the legal engineer spectrum have been sitting in this market for years, undervalued, overlooked, or bastardized into buzzwords. Process engineering is the clearest example. There's a whole cottage industry around it, and plenty of business of law professionals carry real training in it. Valuable training. But most were never given the agency to use it, and candidly, most never got the reps. They know the discipline academically and haven't done the work in a meaningful way, inside live matters, with partners watching. This means market already carries some baseline. It's dated and it's been left unnurtured, but it's a far better starting position than zero, and firms keep mistaking it for zero. And if you've ever watched someone walk into a firm and actually make change happen, grow a practice, contribute to revenue, treat partners as customers rather than obstacles, you already know this profile.

Second, what follows isn't a thought experiment. It comes from the field: my advisory work with firm leadership on exactly this problem, and the course I'm building at IU Maurer this semester, in partnership with Ice Miller, to turn law students into forward deployed legal engineers. I've been testing this profile against real people doing real work. It holds.

Now decompose the need into what the person, or people, must actually hold:

Six components. Three are judgment about the work.

Two are judgment about the system.

The sixth bridges them.

Know this. Almost nobody scores max on all six. It is exceedingly rare. Not the relabeled KM director, not your best senior associate, and not the vendor's engineer either, who typically craters on the legal and commercial axes. The full hexagon doesn't typically exist in one person at market scale very often. If you find one, get the, pay them handsomely, and give them meaningful work and agency. That's why Norm's posting filled in a blink, why Kirkland pays Palantir instead of a recruiter, and why Harvey pays sales comp for legal credibility. Even the deepest pockets in the market lost the unicorn hunt, and they've stopped pretending otherwise.

Which means you should stop looking for a hire. Instead, what covers the hexagon is a small team, three or four spiky people whose shapes overlay into the full profile. A lawyer brings the legal judgment spike. A business of law professional brings the workflow and commercial spikes. Someone grows the technical and evidence axes on real work. Your firm doesn't need to win a talent war it can't win. It needs to compose a pod from people who, in most firms, already work there, quietly using these tools ungoverned because nobody gave the capability a name, a mandate, or a career path. The vendors found your people. The question is whether you find them first.

How to design the role from the need outward

Five steps, in order. Notice the job posting comes last.

And when the next definitive guide to the legal engineer lands in your inbox, run the byline check before you forward it to your executive committee. Who wrote it, and where does the role sit in their business model? Their definition served their need. Yours should serve yours.

The noise around this role is loud and getting louder. The signal is tough to find. The title is a costume. The need is real. The firms that get this right won't be the ones that found a unicorn; they'll be the ones that named their own pain, scored their own people, and built the capability inside, where it compounds.

Everyone else is building shallow, for show. Build deep, for keeps.

Talk soon again, Josh

Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. Not rough transcription you have to clean up. Actual polished text — ready for email, Slack, or any app.

Speak the way you think. Go on tangents. Change your mind mid-sentence. Flow strips the filler, fixes the grammar, and gives you text that reads like you spent five minutes writing it.

89% of messages sent with zero edits. Millions of professionals use Flow daily, including teams at OpenAI, Vercel, and Clay. Works on Mac, Windows, and iPhone.

Start flowing free

To read previous editions, click here.

DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please /be careful and do your own research.