Legaltech firms can't build scalable workflows with AI tools like Harvey
10 posts across 2 communities in the last 90 days; 90% from r/legaltech.
“Walmart was just a half baked example. Take any major tech company- is the technology suite they are offering cheaper now than it was before AI? No. Lawyers will spend the same amount of time working on projects. If individual times on a task are reduced the time will be made up…”
“My issue with Harvey isn’t whether it’s useful today. It’s whether it’s worth paying a premium for. It’s getting squeezed from both directions. OpenAI, Anthropic, and Google are building the actual engines, and every improvement in research, agents, document analysis, and…”
“Agree but guess my point is, have any of these tools aside from H&L owned a single workflow successfully? [re: No moats left in legaltech]”
Legaltech companies struggle to create scalable, user-authenticated workflows using AI tools like Harvey or Legora. Challenges include single-tenant data management, handling finicky lawyer preferences, and integrating with third-party platforms like iManage.
SOURCES (27)
workflows aren't moats buddy. [re: No moats left in legaltech]
“Agree but guess my point is, have any of these tools aside from H&L owned a single workflow successfully? [re: No moats left in legaltech]”
“Walmart was just a half baked example. Take any major tech company- is the technology suite they are offering cheaper now than it was before AI? No. Lawyers will spend the same amount of time working on projects. If individual times on a task are reduced the time will be made up for elsewhere. People don’t pay for tasks to be done, people pay for entire projects which are made up of a number of tasks, some of which AI can help with and some of which AI cannot. That is of course assuming the firm”
Same from the lawyer side. The takeaway is "only if youre really smart"
“Yeah, but I’m not sure that’s going to work. Harvey is consolidating on the same front. Also, caselaw is absolutely not a moat. Otherwise, midpage would’ve been bought by now.”
“Web and internal knowledge base searching is commoditized. PracticalLaw still holds some value but I don’t understand why legal tech challengers aren’t developing that vertical. They have SO MUCH cash”
“We're primarily litigation so lots of variability. I can totally see flat fee being easier to handle with cookie cutter type casework.”
“My issue with Harvey isn’t whether it’s useful today. It’s whether it’s worth paying a premium for. It’s getting squeezed from both directions. OpenAI, Anthropic, and Google are building the actual engines, and every improvement in research, agents, document analysis, and workflows eats into what Harvey adds on top. Meanwhile, lawyers are becoming more AI-literate, and the models are making it increasingly easy to build bespoke tools around your own practice,l and clients. If the underlying tech”
“My issue with Harvey isn’t whether it’s useful today. It’s whether it’s worth paying a premium for. It’s getting squeezed from both directions. OpenAI, Anthropic, and Google are building the actual engines, and every improvement in research, agents, document analysis, and workflows eats into what Harvey adds on top. Meanwhile, lawyers are becoming more AI-literate, and the models are making it increasingly easy to build bespoke tools around your own practice,l and clients. If the underlying tech”
“I agree that in many cases the Harvey, Legoland, amd.other legal tech do not offer enough value relative to leverage frontier model APIs and.ypur own custom apps and agents. Building in house os getting faster and cheaper, amd.ypu can customize and optimize the product to your practice. In.my opinion, lawyers are the original coders. Its time we leverage modern technology to rdesign pur industry.”
“What jurisdictions are ok with Claude Team? AFAIK Team doesn't provide for Zero Data Retention. Only Claude Enterprise.”
“We just went through a whole process trialing Harvey, legora, and cocounsel. We decided, for a number of reasons, that it would be better for us to build our own practice area focused tool for a lot cheaper. Given I keep seeing posts like this, I may write up our reasoning on this subreddit. Might be beneficial to some.”
“Harvey is just a harness that offers you things you can create yourself, plus the Lexis connection. Remove Lexis and, imho, there’s no real viable business case for Harvey”
“We actually showed that training a model on our benchmark generalizes to other 3rd party benchmarks (mercor agents, legalbench, crosby's redline bench) - our model wasn't trained on those 3rd party datasets so I think this actually shows something meaningful about the benchmark.”
“I waiting with bated breath for release of their benchmark results on their own benchmark…”
“Harvey says it can weigh in here after the benchmarks are complete. I appreciate your question, though, because it gets to a key difference: it's one thing to reduce Harvey's own costs by routing more work through in-house models, and it's another thing to pass those savings along to customers. Stay tuned!”
Tenet = donwloaded an open-source model and added another thin layer.
“First, things like user authentication and connecting to iManage are extremely basic. Virtually every single legaltech solution can do this. The vast majority offer single tenancy as well (and it's highly debatable whether you actually need that). If you think Harvey optimizes every part of the data pipeline, every edge case, with rivals, pulls the right context, and whatever else you were going to say, you've drunk some serious kool-aid. And that is exactly how they got to where they ar”
“"real privacy controls" I would be careful with this with any of the SaaS legaltech vendors out there. I have not seen a great deal of depth here. Just because it is claimed in the marketing materials, does not make it true.”
“I am very satisfied Legora user. It really helps me drafting and (proof)reading documents. Also it’s legal search abilities have improved. The Outlook plugin is amazing. I must admit that I haven’t tried any other tool. Only concern we have is about their price increase, which will probably be a limiting factor in its usage.”
“Has anyone compared Harvey and Legora in a side by side comparison? We have the opportunity to go with both, but each attorney will have one or the other. Do you have advice on whether Harvey or Legora matches up with junior attorneys and more senior level attorneys? We are in-house in the healthcare space. TIA”
“Used Legora and it completely changed the way my tasks are done, how projects are organized, how documents are stored. I think they’re also different because they are having the lawyers approve ‘plans’ and review work with direct citations and explanations behind their reasoning and results. They also integrate directly into Word, imanage and I believe your Outlook email, which I heard from friends who use Harvey, is a feature it lacks. It’s awesome to just ask AI whether local counsel sent X or”
“That’s fair. Take any workflow. Don’t care what kind. Can you build it using Claude code, yes. Now add user authentication for hundreds of people, single tenant data management, inside say Azure, that then optimizes every part of the data pipeline, every edge case within a practice workflow, with Evals, the finicky lawyers who each have different taste, then pulling context from the right third party data platform like imanage or whatever. I can keep going. If your point is I can drop a document”
“Biggest reason to use Harvey/Legora over Claude - it's anyone's guess who will have the best AI models in 5 years. OpenAI had the lead for a while, then Anthropic leapfrogged them. And open-source models are getting better and better. If you go with Claude, you're locked into using Anthropic's models and it's going to be a pain if they fall behind another provider. With Harvey/Legora, you can use any model and switch back and forth without any disruption to the end-user exper”