Few people love this industry more than I do, so it pains me to say this: I’m disappointed in us.
We should be championing technology that can be a transformative force for good in legal. Instead, we too often reveal fragile egos and fall prey to tall poppy syndrome*, hating on companies that we perceive to have grown too fast, become too successful, or that have not played by "our" rules.
A number of articles and posts recently and over the past year have called attention to the controversy surrounding Harvey, one of the most successful companies ever to hit legal tech. To call it a “controversy” is a bit of a misnomer, because there is almost nothing controversial about Harvey except its phenomenal success, and the fact that for the first year of its existence it flew under the radar and did not perform the usual circuit of legal tech podcasts, webinars, conferences, social media, and blogs that we have come to expect of our own. I called Harvey out myself in 2023 for having no website to speak of, and for neglecting to explain its product except to the customers it was actively selling to. I have since come full circle on Harvey though, and the more I hear and read from Harvey’s detractors, the more disappointed I feel about industry players who appear to have forgotten what we stand for.
Legal tech geeks, a reminder: This industry isn’t about us. It’s about legal professionals and it’s about consumers of legal services. What makes our jobs exciting and interesting is the proximity we have to disruption: we are cheerleaders for change. Or we should be. Instead, I have watched since the Atrium days as our pundits and tech bros and money men sit on the sidelines and cheer when a provider seen as too big for its britches falls.
This kind of schadenfreude does not become us, nor is it helpful. We risk losing credibility as an industry if we support only those vendors who play by the arbitrary rules we have established – that you must seek funding from certain places, sell to certain people, and speak to specific people before daring to be successful.
It makes me wonder whether perhaps, just as generative AI marks a real shift in the way we practice law, a unicorn like Harvey with its mainstream Silicon Valley funding and its Fortune 500 following marks a shift in the relevance of legal tech as a niche ecosystem. Because all signs point to this: Harvey is here to stay. Harvey criticism stems largely from other vendors, or from the legal tech fringes. The lawyers who use it are fans. Watching from the sidelines the rise of what Dr. Megan Ma of the Stanford Center for Legal Innovation has called “the Harvey Associate”, young lawyers who cannot practice law without this technology, should be a joy for all of us who have waited years for exactly this kind of transformation. If instead it irritates you, you need to ask yourself why and whether what you truly seek is positive change.
Let’s take an objective look at the criticisms lodged at Harvey.
Fiction v Fact in What We’ve Heard About Harvey
Scuttlebutt about Harvey takes several forms. Though I do not have the inside scoop on all of these, an exploration of the most prevalent rumors shows that we’ve been far too harsh.
- Skepticism about the tech: “Harvey is no more than a wrapper / ChatGPT”
Most of us may have been in the dark about Harvey tech in 2023, when Harvey wasn’t talking about it. In 2024, though, Harvey has been active in publishing blogs that explain in detail how the platform operates (follow along on LinkedIn if you’ve missed these).
To suggest that Harvey is anything remotely close to a “GPT-wrapper” or “essentially ChatGPT” (both of which criticisms I’ve heard) is ludicrous. It also makes no sense. Earlier this year Harvey acquired Mirage, a custom large language model (LLM) designer, and established a partnership with Mistral, a French provider of open-source language models. Anyone paying attention to industry news will be aware that Harvey has more going on under the hood than ChatGPT.
Another reason this rumor seems ill-spirited: we all know how much funding this company has raised. More than any other company in legal tech, they have the resources to pour into developing the product. Unsurprisingly, Harvey has recruited top development talent and is building best in class technology for the legal market. As a result, Harvey is one of the most sophisticated legal AI platforms from a technical perspective, with the following characteristics:
-
- Multi-model and model agnostic: the system is powered by a range of commercial grade, open source and custom language models, with the ability to continuously update the models used as appropriate to optimize performance.
- The commercial models it currently uses most are GPT-4o and OpenAI o1.
- Harvey also has a custom LLM that is trained on a very large data set of legal documents, as well as human reinforcement learning by attorneys.
- Harvey leverages Mistral’s open-source models.
- Harvey has built a number of jurisdiction specific research models.
- Harvey constantly evaluates different models for their speed, context window, and ability to be used in different tasks, pulling in additional suitable models for the right tasks as the landscape evolves.
- Complex model calls: Harvey chains together model calls and decision trees to produce outputs. If you run a query in Microsoft CoPilot or in ChatGPT, that will involve a single model call to the foundation model before it returns an answer. By contrast, when you run a query in Harvey, there’s a chain of 30-150 model calls in most instances. Some Harvey tools may chain over 1200 model calls to produce a single answer, and to ensure the reliability and relevance of that answer. For every answer produced by Harvey, there are quality assurance model calls that read the answer before the end user does, to ensure it is correct and optimal. And yet, Harvey is still able to operate rapidly. These chained model calls allow for far greater legal specificity in Harvey’s analysis than most other legal tools. It’s a far cry from ChatGPT.
- Agents: For Harvey’s workflow product, it deploys custom developed agentic AI combined with model calls to build out end-to-end functions that support lawyers from the beginning to the end of specific tasks.
- Multi-model and model agnostic: the system is powered by a range of commercial grade, open source and custom language models, with the ability to continuously update the models used as appropriate to optimize performance.
- Limited use cases: “Harvey just summarizes”
I’ve heard people refer to Harvey’s applicability as shallow, insinuating that it performs only surface-level tasks “like summarization,” similar to ChatGPT. First, it’s important to recognize that producing a good legal summary is no trivial feat. Producing a high quality, reliable summary of complex legal documents is challenging, and ChatGPT would not be able to perform this use case at an adequate level for use in legal practice.
In fact, Harvey addresses many legal-specific pain-points. We published a Use Case Matrix on LTH Premium several weeks ago, comparing the use cases addressed by six of the main legal AI providers. Harvey tied for first place in the use cases it addresses (about 20 were listed), and more workflows and use cases are being added regularly. For those who don’t have access to LTH Premium, some of the use cases covered by Harvey include:
-
- Document Q&A
- Drafting from scratch and from precedent (also in Word)
- Contract review
- Data extraction (at scale) and due diligence
- Red flag review
- EDGAR research
- Multi-jurisdictional tax research
- Redlining for negotiation
- Clause suggestions
- Chronologies
- Deposition / Cross-examination outlines
- Targeted search (such as for eDiscovery)
- Transcript management
- Digital translation (multi-language)
As Harvey workflows continue to be built out, lawyers will see even more specific use cases and pain-points addressed by the system.
But also, this: Harvey’s investors include the biggest VC names in Silicon Valley. Detractors should be asking why those investors would be champing at the bit to invest if Harvey does nothing useful. We do not have better insight into Harvey’s worth than the most successful venture capitalists in the world. Sequoia and Kleiner Perkins absolutely do their research before investing millions of dollars in a company.
- Hiring the wrong people: Harvey hires lawyers and go-to-market, not developers
Another criticism I’ve heard is that Harvey has spent its money on hiring go-to-market professionals and lawyers rather than developers and engineers. See 1 above for discussion of Harvey’s deep investment in development talent.
In terms of lawyers, while it is true that Harvey has hired a large number of them, there are good reasons for it. Harvey’s go-to-market is driven by lawyers, and lawyers are also used in product development. In a recent interview with Harvey’s Head of Strategic Business Development, Jake Weiner, he said that a lot of lawyers are hired to “work one-to-one with engineers and the product team.” Subject matter experts are assigned to work alongside engineers, helping to bake in legal expertise as the product is developed and helping to test and evaluate how well the product is actually working.
- Poor renewals or sales: “Harvey is seeing a lot of churn”
Despite clear evidence to the contrary, rumors still abound that Harvey is suffering from poor sales performance, or that it is about to go under because customers are not renewing.
As with any industry, legal is competitive. I’m aware from conversations I’ve had with law firm leaders just how many are using Harvey. Those that have good traction with the platform tend to be fairly quiet about it, since this is seen as a competitive advantage. These are not firms involved in innovation theater. They are more interested in getting work done than in press releases.
Take a look at the Harvey website to see the pedigree of its customers. Though precise revenue is unknown, Harvey has shot well past $40M in annual recurring revenue, and it is only two years old. That is pretty phenomenal growth. In spite of rumors about churn, the firms I’ve spoken to are not just renewing but increasing their Harvey spend.
- Trouble in paradise: Harvey is on the brink of failure because two BigLaw partners left
Yes, two BigLaw partners who joined Harvey early on recently left the company, as was reported in the legal tech press. In the absence of any other evidence that Harvey is in trouble, why would we assume that the departure of 2 out of 200 employees is somehow ominous?
Let’s face it: if this was not a story about Harvey, but about another vendor, our schadenfreude would reside elsewhere. We in legal tech are open about our skepticism of law firm partners’ ability to rapidly transform the way they practice and operate. Here were two BigLaw partners leaving extremely well-paid jobs to join a tech startup. While I laud (and understand) their enthusiasm, it is no surprise to me at all that the scrappy environment of a startup did not ultimately suit them. Are we really surprised that they went back to familiar, stable law firm jobs? It was an interesting experiment. The fact that these BigLaw partners did not stay long at a startup probably reinforces a few of our long-held beliefs. That should be our take on it.
- Arrogance of founders / team: “The Harvey team is pushy”
This is an interesting one, and though I have never been a customer of Harvey’s, I can say that for every critic of the Harvey sales approach, I encounter more fans of it. The innovation leaders I know whose firms have licensed Harvey tell me that there is a phenomenal amount of support from the Harvey team.
I’ve also attended a social event with dozens of Harvey customers from multiple regions. I asked a lot of questions that evening about Harvey and heard nothing but glowing reports.
- Harvey seeks to replace lawyers
Less rumor than speculation, some pundits have said that Harvey’s name (presumed to stem from Harvey Specter of Suits) suggests the intention of its founders is to replace lawyers. Others have suggested that the tasks taken on by Harvey are too sophisticated and should not be automated.
The Harvey website clearly celebrates lawyers, though, and the language used in all Harvey messaging makes it plain that Harvey seeks to augment and uplift lawyers rather than replace them. Murmurs about the name seem trite in an industry where literally hundreds of providers include the anchor word “Lex” in their product branding.
Your Call to Action
Look, I’m not saying Harvey is perfect. But anyone throwing shade on Harvey or any other unusually innovative and successful company in legal should check first whether that shade is warranted.
It strikes me that we are in danger of falling prey to that most English of attributes, *tall poppy syndrome, a social phenomenon that stems from resentfulness for the success or achievements of others. Instead of celebrating success stories, there is a tendency to diminish them. Like a tall poppy in the field of legal tech, Harvey is such a stand-out that it’s tempting to pull it down rather than lift it up as one of “our” successes, especially because we didn’t feel, at first, that Harvey was really “one of ours.”
We need to be better than that. If what we want as an industry is to drive positive change, we should feel inspired by Harvey, not threatened. And for those of us that have noticed Harvey’s increasing transparency, its growing presence at legal tech gatherings, and its widening moat of loyal customers, let’s open our arms and invite them in rather than continuing to cast stones.
Because one day there will be another Harvey, another new standout success story, and we need to have learned from our mistakes.
