Conversation
Highlights

How Eve Reached Billion-Dollar Valuation by Targeting Labor-Constrained Markets Over Budget-Constrained Ones

At a plaintiff attorney conference in early 2024, Jayanth Madheswaran asked the room a simple question: how many people use AI at work?

Two hands went up out of a hundred attendees.

Fast forward to late 2024 at the CAOC conference, and the same question yielded a dramatically different result. “Literally everyone, all but like two people raised their hand,” Jayanth recalled in a recent episode of Unicorn Builders.

This wasn’t gradual adoption. This was faster than mobile internet penetration in a traditionally risk-averse industry. And it revealed something critical about the market Eve was building for: plaintiff attorneys weren’t experimenting with AI—they were already underwater and desperate for capacity solutions.

The NLP Data Efficiency Bet That Preceded LLMs

Jayanth didn’t target legal tech deliberately. After leaving Lightspeed in 2018-2019, he and co-founders Matt and David identified an exponential trend in natural language processing that most founders missed.

“The amount of data you needed to train a model was dropping by an order of magnitude,” Jayanth explained. Training simple classifiers that once required millions of records now needed just thousands, then hundreds, potentially just ten examples. “This was even before LLMs.”

They built Butler for document extraction—contracts specifically, where every document is structurally different but contains the same extractable data points (amounts, start dates, end dates). “The hard part about contracts is the old OCR models don’t work well. Every single contract is so different, but they all have the same information you may want to pull out,” Jayanth noted.

Then LLMs arrived and compressed years of roadmap into months. “Suddenly you can go so much deeper into the workflows than you could before for a fraction of the engineering cost.”

Mid-2023, they saw it: what started as document extraction could now automate entire case workflows from client intake through settlement. They launched Eve in early 2024.

Why Plaintiff Law Economics Create Natural AI Adoption Incentives

Most founders evaluating legal tech see a single market. Jayanth saw two fundamentally different businesses with inverted incentive structures.

Plaintiff firms only collect fees on contingency when they win. “When they represent you, they’re kind of going out on a limb and fronting all the legal costs and other costs associated with representing you themselves in hopes of doing really good job for you and then sharing in a percentage of your winnings,” Jayanth explained.

Defense firms bill by the hour. “They might package it similarly, which is you get sued, they’ll say, hey, I’ll take you up to dismissing a complaint… they would estimate some hundred thousand dollars for it, but it’ll be billed hourly. So there’s no real cap in most cases.”

This creates structural capacity constraints on the plaintiff side. Each additional case represents unknown hours of work—a wrongful termination case might settle in weeks or litigate for years. Scaling by hiring creates financial risk because headcount is fixed cost against variable, uncertain revenue.

Meanwhile, plaintiff firms face massive resource asymmetry. “Defense is funded by companies with a lot bigger budgets and they hire big law firms which a lot more people can work on it. And the plaintiff side you have, you know, sometimes it’s a two person shop defending you.”

The result: severe unmet legal demand. “You hire a contractor and they didn’t do something and they owe you 500 bucks—in most cases, it’s not going to make sense today to go get a lawyer for that.”

This is why adoption accelerated so rapidly. AI doesn’t just improve efficiency for plaintiff attorneys—it breaks their capacity ceiling entirely. “On a case level, you’re talking hundreds of hours saved,” Jayanth said.

One Atlanta firm maintained headcount while adding enough capacity to take pro bono cases under their previous $5,000 minimum threshold. “Now they’re able to go represent them and then change the lives of some individuals forever.”

Pricing to Existing Budget Line Items, Not Creating New Procurement Categories

Eve charges per case, not per seat. This wasn’t innovation—it was pattern matching.

“We typically charge per case. So per matter that you take on, which indirectly represents like a client. Each new client you represent, it’s some amount of dollars per client,” Jayanth explained.

Law firms already have this mental model. “Even today in our market, people are already used to paying external vendors on a per case basis for such help. So for example, if you have some labor and employment case that needs some intimate medical knowledge, you actually would go and hire out someone with a medical degree to serve as an expert witness potentially.”

The category positioning is equally deliberate. Today, Eve is workflow automation. But Jayanth sees the evolution clearly: “Most likely I think we’re going to go through this shift between software as a service into like service as software.”

As agent accuracy improves, the product resembles hiring specialized employees. “The end state for agent is like some sort of employee. It’s able to handle the same type of work that an employee that’s trained on a particular task would do. And then now you’re back to accuracy it reports again.”

The line item Eve ultimately replaces? Those external vendors charging per-case fees. “That line item is most likely what Eve is going to replace.”

The procurement path of least resistance isn’t creating a new software category—it’s replacing existing service spend with software that delivers the same outcome at the same unit economics.

Why Physical Presence Compounds Referral Velocity in Network-Driven Markets

While most B2B companies optimize digital channels, Eve built heavy field presence at plaintiff attorney conferences. The reasoning goes deeper than “relationship building.”

“Lawyers, fundamentally they’re a trust based ecosystem and network based ecosystem,” Jayanth explained. “They are used to referring cases to lawyers in their network. They get cases that they can’t take on, they try to refer it over to someone else so the client gets seen.”

This referral infrastructure is how the market actually operates. Attorneys with specialized expertise receive referrals from generalists. Overburdened firms refer overflow work to trusted colleagues.

Product recommendations flow through these same channels—but only after establishing credibility. “One of the things that’s really important for us is how do we show up for them,” Jayanth noted. “We actually have a pretty heavy field presence. We go to a lot of these conferences and participate as much as we can in the same types of things that would keep a plaintiff attorney up at night when it comes to running their business.”

The daily usage creates organic word-of-mouth. “One nice part is the product really does work and it gets used every single day… that’s led to them talking about it and referring new clients over as well. So good chunk of our business actually. Word of mouth growth.”

The field investment compounds because high-engagement products become conversation topics within established networks. Digital outbound can’t replicate this—attorneys now detect AI-generated emails and ignore them.

Embedding Operators to Build Proprietary Workflow Context

Eve employs actual plaintiff attorneys in-house. Not as advisors—as product builders determining where AI should and shouldn’t penetrate workflows.

“We’ve hired really excellent plaintiff attorneys to help be on the forefront of what are the right places to use AI, what are the places we should not be using it, what are the places we should be cautious, what are the places we should be accelerating and testing the boundaries,” Jayanth explained.

This builds competitive moat through proprietary context. Jayanth referenced a paper by Andreessen Horowitz partner David Haber titled “Context is King” to explain the necessity: “You need to have such a deep, intimate understanding of the technical side mixed with the product side mixed with actually one of the workflows are doing what are the gotchas in that vertical and you have to add expertise in.”

The gotchas—edge cases, compliance constraints, workflow exceptions—only surface through operational experience. “We have to create our own engine to find that out. And as a result we now have that context that we can then use to make better and better products.”

Competitors can hire consultants or conduct user research. They can’t replicate the institutional knowledge that comes from having former operators embedded daily in product and engineering decisions.

Tracking Founder Removal Rate as the Sales Scalability Metric

Eve transitioned out of founder-led sales faster than typical B2B companies. The trigger wasn’t revenue milestones or team size—it was a specific behavioral shift.

“There’s a pretty big difference between when founders are the ones closing the deal versus the reps themselves start closing the deals. It’s a pretty marked shift,” Jayanth explained.

He stayed involved in every deal but tracked what closed without his participation. When reps consistently closed independently, the motion became scalable. “Then thankfully after that things become more mathematical. You start looking at patterns and making incremental improvements.”

The first two sales hires took deliberate vetting—20 to 30 candidates before hiring Ryan and Christian. “Being the first sales rep at a company is never easy. You have a high probability of failure, both yourself individually and the company.”

He looked for specific traits: relationship-building ability, understanding of buyer psychology, and market intuition. Early reps need to read customers, understand decision-making processes, and adapt without extensive process.

The process philosophy was minimal constraints with maximum latitude: “Make sure you get paid when you win deals” and “make sure we’re actually being truthful and selling what we have and being very clear where we don’t have it.”

The second rule matters critically in network-driven markets. “I think that type of selling motion is one that is very common in early stage startups with lawyers. It burns a lot of trust and you don’t want to do that. You want to actually have a partnership.”

Misrepresenting capabilities doesn’t just lose one deal—it poisons referral networks permanently.

Qualifying Early Customers on Future-State Vision Before Product Features

Early customer selection wasn’t about who had the biggest pain or largest budget. It was about worldview alignment.

“What we want to do is meet them and actually talk to them first about, okay, where do you actually see are the biggest challenges in your law firm? And where do you see AI helping? And what do you see the future? Like, are AI agents going to be running around all over the place? Or are people going to be used? Where are people going to be used?” Jayanth explained.

“You want to get alignment on that vision first. And that’s how we found our early customers.”

These design partners weren’t just adopting a product—they were co-creating category positioning. “We coined the term like AI native law firms. And that was in conjunction with a lot of our early customers.”

The second wave required different engagement: “Once you articulate, here’s how you should be thinking about your plaintiff law firm, here’s how you should be thinking about using AI for these workflows. Here are the features that matter… When they see it, they know it.”

Vision-aligned customers tolerate product gaps because they’re building alongside you. They provide the insights that become your category education content. They become reference customers because they’re genuinely enthusiastic about the future you’re creating together.

Feature-focused customers churn when the roadmap shifts or competitors match capabilities.

Hypergrowth Through Product Velocity, Not Just Headcount

Eve scaled from 13 to 120+ people in twelve months while launching six products in eight weeks. “Even having been at Facebook and having been at Rubrik, this is just another level,” Jayanth noted.

The challenge isn’t building features—it’s maintaining alignment at scale. “The entirety of the company does not fit in my brain right anymore. There’s a number of people doing amazing work that I only see little snippets of now.”

His current focus: “How do you give the feedback to every single individual such that they’re all working collaboratively in the most effective way possible.”

It’s the universal founder challenge at hypergrowth—information flow breaks down before org structure does. The solution isn’t complicated: “There’s all sorts of management techniques made for this.”

But it requires constant attention as the constraint that limits velocity shifts from building to coordinating.

Two months ago, Eve reached billion-dollar valuation. Jayanth missed his own celebration—he and his co-founder had an urgent opportunity. His reaction wasn’t triumph but “one of gratefulness.”

“We would not have gone here without a lot of the early bets our customers took. Early on, people had to wait time before we actually delivered on our product. And that’s kind of crazy. And that’s a huge trust that they placed on us and I didn’t want to let that down.”

The trust was justified. Most customers maintained headcount while dramatically expanding caseloads—either taking more cases at their quality bar or going deeper on complex litigation requiring extensive discovery work that would have “previously required hundreds of paralegals.”

The capacity didn’t just grow revenue. It changed outcomes for individuals who previously couldn’t access legal representation—the $500 contractor dispute, the employment case under $5,000. Problems that were economically impossible to solve now have solutions.

That’s the real measure of product-market fit in labor-constrained markets: not just efficiency gains, but entirely new outcomes that were structurally impossible before.