Everybody in B2B AI is rediscovering forward deployed engineers right now. Palantir ran the model for two decades, OpenAI built a team, and everyone has figured out agents don’t deploy themselves effectively. Not well at least, not yet at least.
Harvey runs two versions of FDEs, and the interesting one isn’t the classic technical FDE pod. It’s the lawyers.
Anique Drumright, Harvey’s Chief Product Officer, walked through both on stage at SaaStr AI. Harvey’s forward deployed pods are mixed teams: a product manager, one or two actual lawyers, and software engineers, working bespoke for the customers that need it. Not everybody gets one.
What everybody gets is a legal engineer. Harvey has roughly 180 of them working with law firms and in-house teams, and they’re former practicing attorneys. Anique’s number on stage was 8 to 10 years of practice for most of the team. They’re in every deployment, not just the big ones.
For a company selling into more than 60% of the Am Law 100, with 1,400+ customers in 60 countries and over 100,000 lawyers on the platform, that’s a deliberate and expensive structural choice. Five things worth taking from it:
- The domain expert is staffed on the account , not advising from a distance.
- The function splits three ways , pre-sales, post-sales and custom solutions, and Harvey publishes the distinction.
- It costs $220,000 to $320,000 OTE , on a 75/25 split, plus equity. That’s public on their own careers page.
- Legal engineers shape the roadmap , which is what makes the cost defensible rather than pure services drag.
- Harvey is now certifying legal engineers who don’t work at Harvey , which is the part almost nobody has noticed.
#1. The person in the room practiced the customer’s job for a decade. This is key #
The standard version of forward deployed engineering is a strong engineer who learns your customer’s domain on the job. They get good at it in six to nine months. That works when the domain is knowable in six to nine months.
Legal isn’t. A litigation partner can tell in one meeting whether the person across the table has ever actually run a matter. So Harvey skips the ramp and hires people who already did the work. Their own words for the team, repeated across every job posting and their careers site: lawyers who left practice, and a team composed of former practicing attorneys. The posted bar is a JD or equivalent plus at least three years practicing at a top-tier firm, in-house, or advising a corporate legal team. Anique’s number on stage was 8 to 10 years for most of the team.
What that buys, in her framing, is credibility in the discovery conversation. The legal engineer asks the questions a practicing lawyer would ask, which surfaces the workflows a software person wouldn’t know to ask about, and the adoption barriers a customer wouldn’t volunteer to a vendor.
#2. Harvey splits the function three ways, and publishes the split #
Most companies have one blurry role covering pre-sales, onboarding and expansion. Harvey segments legal engineering into three teams:
- Legal Engineers (pre-sales). They lead discovery and demos alongside AEs. Harvey’s own language for it: solutions architects secure the technical win, legal engineers secure the legal win.
- Legal Engineers, Product Specialists (post-sales). Onboarding, enablement, adoption, utilization, and support for renewal and expansion, working with CSMs and AEs.
- Legal Engineers, Custom Solutions (pre- and post-sales). The bespoke build work. They meet with practice group heads and general counsel, learn how a process runs today, then design the agents and playbooks themselves using natural language prompting. A lawyer builds it. No engineer writes code for it.
The reason the split matters: it makes the role hireable and it makes the cost attributable. A pre-sales legal engineer is a sales cost. A post-sales product specialist is a retention and expansion cost. Custom solutions is a services cost. Most companies bundle all three into “solutions” and then can’t explain their gross margin.
The post-sales version goes deep enough that Harvey describes it as sitting down with a full practice group to build agents on the firm’s real matters, not on training exercises.
#3. The public number: $220,000 to $320,000 OTE, 75/25 split, plus equity #
Harvey posts the range on its own careers page for the Product Specialist role. Three things to notice.
First, it’s variable-comp. A 75/25 split means this person carries a number, which is unusual for a post-sales domain expert and tells you Harvey treats adoption as a revenue function rather than a support function.
Second, at that comp you are competing with what a mid-level associate makes, which is exactly the point. You cannot pay a solutions engineer salary and hire a practicing lawyer.
Third, run the math on 180 of them. Even before equity, you’re carrying a line item in the tens of millions to put a lawyer in every deployment.
Harvey is doubling down on it anyway. When they raised at an $11B valuation in March, the stated use of proceeds was to expand the agents customers run on Harvey and to grow the embedded legal engineering teams supporting them globally. The round is partly funding headcount most B2B companies would call services.
#4. Every deployment gets a legal engineer. Only some get an FDE pod. #
This is the distinction I pushed Anique on, and it’s the one most people miss.
The forward deployed pod is the mixed team: a product manager, one or two actual lawyers, and software engineers, working bespoke on that customer’s problems. Anique was direct that not every customer needs one and not every customer gets one.
Legal engineers are in every deployment, always, because Harvey’s position is that a subject matter expert is what lets a customer get the full value of the platform. The expensive engineering resource is rationed. The domain expert never is.
Most agent companies have this backwards. They ration the domain expertise and scale the generic implementation resource, then wonder why adoption stalls at the pilot.
#5. The legal engineers write the roadmap back into the product #
Legal engineers shape the product roadmap so every release is grounded in what the profession actually needs. They’re athe trust layer across the customer lifecycle, from first pilot to the most complex custom workflow.
There are now more than 25,000 custom agents running on Harvey across M&A, due diligence, contract drafting and document review. Legal engineers work alongside customers to build and improve them.
That’s what keeps this from being services drag. A hundred and eighty former practicing attorneys sitting inside real matters at the best firms in the world is the highest-quality product research operation in legal, and it bills.
Compare that to how most B2B companies get product input: a CSM relays a feature request, a PM interprets it, and three layers of translation later you build something adjacent to what the customer meant. Harvey’s translation layer is a person who used to do the job.
#6. Harvey is certifying legal engineers who don’t work at Harvey #
Harvey Academy now runs a Certified Legal Engineer path. Self-paced, open, with a shareable badge. Harvey’s pitch for it draws on their experience supporting customers through more than 1,000 AI rollouts, and their framing is that legal engineering is a new profession combining legal judgment and technical fluency, still taking shape, and that the people doing it now get to define it.
That’s a business decision more than a marketing program. Harvey cannot hire enough practicing lawyers to put one inside every legal organization in the world, so they’re turning the role into a credential that people at customers, at firms and at other legal tech companies can earn.
he buyer’s own staff gets trained on Harvey’s definition of the job. The category’s vocabulary is Harvey’s vocabulary. And the constraint on their deployment model stops being their own headcount.
Harvey’s own adoption number, from the same page: AI adoption across law firms and legal departments went from 14% to 43% in two years. The certification is a bet on where the next 57 points come from and who guides it.
#7. What it would cost you to copy this #
The model isn’t free, so here’s the accounting.
- Gross margin. Every deployment carrying a $250K+ human is a real hit. It’s defensible at Harvey’s contract sizes and their category’s willingness to pay, and it may not be at yours. Run it before you commit.
- Hiring pool. Harvey is recruiting from a profession where people leave $400K jobs for mission and equity. If your vertical’s experts aren’t looking to leave, this doesn’t work.
- Ramp and retention. You’re hiring people who spent a decade in a different career. Some of them will miss it.
- The management problem. A former partner is not going to take direction the way a two-years-out solutions engineer does. That’s the same reason they’re valuable.
Most of us need forward deployed engineers and don’t have them #
When you buy Harvey, you get someone who spent a decade doing the work, on top of the software. Very few agent companies can hand every customer a subject matter expert.
Who is the person your customer would trust to redesign their workflow, and are they on your payroll or your customer’s? If you’re counting on the customer to supply that person, your adoption rate is set by whoever they happen to assign. Harvey pays for that person themselves, 180 times over.