Atrium shuts down, laying off 100
techcrunch.com
techcrunch.com
I asked if all of their lawyers are based in the US, and they couldn't give me a straight answer. Eventually, when I said "well, of course they must all be US-licensed lawyers, right?" they said "oh, yeah, they must all be in the US then."
It was not a confidence-inspiring conversation, and I can't say I'm surprised that they're shutting down.
The original concept, as a reminder: https://techcrunch.com/2018/09/10/atrium-legal/
Now it's easy to say that this should've been obvious in hindsight. (Should've been obvious to me, anyway; I cold-mailed Justin and Nick years ago looking to build a specific function within Atrium to address concerns I'd heard on background about the service offering; I never did hear anything back about that.) But that said, I can't help but think the original pitch by Atrium was just too bitter a pill for the most conservative industry in the United States to swallow. Shoot, lawyers are bizarrely resistant to moving to e.g. Signal to preserve attorney-client privilege; given this, I have a hard time thinking of many who would trust ML models with understanding case-law.
As a result, they've often been early, broad adopters of general-purpose office and productivity technology. Sometimes to their eventual detriment overall, by standardizing on an early generation at the expense of later incremental improvements. But at the same time, I think US lawyers as a class are remarkably resistant to whiz-bang pitches of especially law-specific solutions. It's easy to sell a lawyer some technology that their counterparts are using to run circles around them. It's hard to sell a lawyer some technology with a change-the-world, techno-solutionist pep talk.
Anecdotally, I can't say how many times I've attended or watched pitches by tech-focused people, business managers, or technophile lawyers with very little practice experience hocking half-baked solutions held up as replacing or supplanting lawyers in some way. My reliable takeaway from those pitches is that the founders don't know what lawyers do. When I ask about tablet ownership, cloud service adoption, security standards, terms of use for professional ethics requirements, or Lexis terminals in the 60s and 70s, it's usually a big, blank stare.
Whoever can build tooling to help lawyers decipher a 100 page credit agreement in half the time, reliably all the time... will do way better than someone replacing those lawyers with AI-outsource hybrids that can do it in 1/10 the time, but only succeed half the time.
That said, as a fellow lawyer, coder, and legal toolmaker, I'd posit there's more of a middle ground than you laid out:
> It's easy to sell a lawyer some technology that their counterparts are using to run circles around them.
Of course – it should be easy to sell lawyers on established technologies! In fact, one could argue competence REQUIRES lawyers/firms to adopt said technologies. It's not wrong to be disappointed with what's being defined as tech adoption in this thread by industry defenders:
> security standards, word processing, LANs, spreadsheets, aggregating data & metrics, owning an iPad, ...
These are bare minimums and best practices in every other industry, and celebrating attorneys for meeting an extremely low bar is no better than technobabble.
> It's hard to sell a lawyer some technology with a change-the-world, techno-solutionist pep talk.
Stipulating there's a difference between snake oil and actually meaningful legal-specific technologies for practicing attorneys/firms, then yes, certain players in the industry will be more conservative about adopting new innovation. Conversely, a functioning market would also imply the opposite: those who responsibly bet on the next big thing, whether in an operational capacity, to drive costs down, or with the consent of their clients.
Sad to say, that's not the current state of the legal industry. Sure, I'm likely biased since I'm still on the vendor side irrespective of how much we aim to make well-built software with solid UIX focused on the practitioner. Nonetheless, firms remain extremely risk averse, hierarchical, obsessed with profits per partner above all else, and long live the billable hour.
Once again, it's complicated, and no one really has the nuance or all the answers, so we all fall back to a bit of a stereotype which makes that middle ground elusive. Ultimately, however, it's really the industry that suffers, and I wonder if/when it will ever change, what the impetus will be, and if it truly requires another global economic meltdown to shake things up.
Such an interesting statement! What do you mean by it?
Say a new technology allows lawyers to do the same tasks in half the time it used to take them. That sounds like it rhymes with "half the pay for the same work" to me, which might account for some of the new-tech skepticism.
That's my take anyway. I suppose it's similar to the knee-jerk reflex to automation, where you're liable to 'automate yourself our of a job.'
Obviously the legal field is very diverse. I'm sure there are 80 year old lawyers out there still using teletype terminals to do their work because that's what they know. But my experience with larger law firms is that they see technological innovation as a key to their success.
In my experience, there is some correlation between technology and age. But keep in mind that 55 year old lawyers were born in 1965, started coming out of law schools in the late 1980s, and became established in the early Internet era, before the bubble burst.
Part of what you're getting at might be down to how the role of older lawyers tends to change, especially in firms. I would say most older lawyers in transactional fields do far more advisory and strategic work, akin to consulting or even lobbying, than issue research, contract review, or routine drafting. If the senior partner's working pretty old school, but all the junior partners, associates, secretaries, and paralegals under them use their own tools, that's still a pretty newfangled picture, overall.
Part of it may also be certain specialties that deal routinely with government bodies, like court systems and regulators, with their own adoption curves. But don't jump to conclusions there based on standards for consumer technology. The websites may look clunky, and the interfaces may be pretty retro. But the databases involved are often functionally complete, resilient, and workable for the severely outnumbered public servants that use them most.
I think one area that we need to improve on is handling of PDFs. It's clearly the standard for documents of record. However, the fact it can be scanned images or text data is inconvenient.
Well known clients include msft/uber. Bitcoin/cryptocurrency practice is well regarded.
I think being close to a lot of tech clients helps. We get some flack for shunning slack/teams/gchat4business/stride/chime or wtf amazon uses/workplace by fb/etc and relying on “antiquated” email. We definitely prefer to work more in thoughtfully composed emails compared to a fluid team chat. It’s also easier to organize and manage.
However, I’ve been pleased with my colleagues ability to ask relatively smart questions about ML. For example, people are pretty good at responding to ML hand wavey bullshit with specific questions about training data, labeling, features, model architecture, etc.
IMO, we have some younger, eg 40-50, people in leadership positions in my practice group.
Just be advised that if the prospect's head actually explodes, you may face capacity challenges to your engagement letter when you bill.
I actually used to work in their Palo Alto office — back when it belonged to MWE in 2007-2009. The new lobby that Perkins built is much nicer!
I feel like the latter model could have worked as a law firm doing the latter. Like, an actual, ordinary, perfectly milquetoast, law firm, hourly billing and all, except with an R&D division that tries to build workflow efficiencies and uses them to compete on hourly rate. Which, come to think of it, isn't that just what Dentons is doing? So maybe it's just American lawyers that are hopelessly hidebound.
I suspect this was really a tribal play. The real plan wasn't automation, it was to pitch the company as a specialised law firm for techpreneurs that had recognisable tribal tech features - less of the "them" that tech people feel around lawyers, and more of the "us."
The problem with law is that you need the "them" parts to be at least competitive with the rest of the industry, on its own terms.
Not only did that not happen, but the "us" parts made no real difference.
I submit that law firms are actually far more progressive about technology than the tech world assumes, but are also a lot better at sniffing out bullshit software that doesn't provide a real-world benefit.
From talking with lawyer friends, I got the sense my experience was not uncommon.
The problem is that there's a limit to the automation. Human lawyers can't be replaced. Interpretation, strategy, communication, court dates, etc. all require the human factor. No amount of software is going to make the human go any faster here.
I think the much more plausible explanation is that a lot of these alleged 'productivity tools' don't add that much productivity.
As a factual matter, you are quite wrong. Law firms rarely get reprimanded for over-billing, and in fact, the opposite can be true.
When a lawyer bills 2X more than a competitor, they can say that they just did a 2X better job, were more thorough, etc. Lawsuits are rarely repetitive, as the facts and legal principles applied vary widely, so it's very difficult to determine what's the correct amount of time to spend on something. Worse, firms that develop reputations for big bills, can even attract big bet-the-company cases where clients can justify paying top-dollar.
The economic principal behind this is that legal services are not a typical "commodity", so standard market principles are not as effective. If you can turn a legal service into a commodity (such as LegalZoom has some parts of the law), you can be much more effective.
Generally, if you want to sell technology to lawyers, telling them they will be more efficient will not work, they are simply not motivated to spend money for something makes something they are already doing take less time.
If you want to sell to lawyers, the best arguments you have are:
- Increase Quality / Reduce Errors: There are tons of mistakes that lawyers can make, and they are constantly terrified of making them. As a lawyer reputation is everything, and even silly inconsequential mistakes hurt.
- Get More Clients: Small-time lawyers spend up to 50% of their time trying to attract clients. At larger firms, it's less than that, but it's still a constant pressure that they are judged on. They may not be willing to pay for something that makes their work 10% faster, but they will pay for something that gets them 10% more clients.
- Capabilities That Change the Law Itself: Some technology really changes the rules itself. E-discovery allowed firms to look through millions of pages of documents, and to bury their opponents in the same. If you didn't have it, you were at a disadvantage and would lose your case.
On slow days, I don’t want to raise my efficiency that much , I take my time and review more thorougly. But on busy days, where I bill 13+ billable hours for a week, I’d want as much efficiency as possible.
I don't think you're really talking about "efficiency" here. On slow days, you may want to spend more time to do a more thorough job, but you still wouldn't want to be less efficient.
Simple automation example: renaming files into a new format. If there was a tool that could rename a large number of files for you quickly, you would use it on slow days and busy days. There is nothing to be gained to doing it manually.
I think you're referring to tools that speed you up but sacrifice quality. That isn't a pure efficiency tool, it's tool that allows you to adjust balance your time and quality of work.
As a lawyer, if you read and understand a document in 10 minutes and bill an hour of time for it, that's lying and you can lose your bar license over it. However, if you read that same document very slowly and carefully over a full hour, then you are considered a high quality detail-oriented attorney.
If you can and want to bill more to a client, you simply review the matter covering more depth or let the work be done by more senior lawyers who have a better hourly rate.
Big-law collects all the client is ready to spend on legal services. The more you spend, the better service you’ll get. If you suspect foulplay in billables, review the job narratives and raise a complaint.
I am telling you --from experience, not theory-- that many lawyers are resistant to time-saving tools. Not all of them, but many of them. They rarely voice out-loud that their concern is about reduced billables, and will instead make absurd arguments that the time-saving tools decrease quality.
Simple real life example: we were selling a very simple bulk download tool to a law firm. Instead of clicking a link and downloading a document many times, this tool would download many in batch and put them in a zip folder. They raised the objection saying "how do we know it downloaded everything", even though the zip file also had an excel listing all the documents that were downloaded, the source of where they came from, and any errors in downloading them.
Law firm prestige is measured in part by billing rates, which firms almost never (except 2008) lower. If a firm lowered rates for any reason, associates would start getting worried/jumping ship. I'm not saying it's rational, just that the field is prestige-based, and associates would freak if they saw their billing rate drop. It would be seen as a harbinger of lower salaries or lousy bonuses on the horizon, no matter how the firm spun it.
Kinda difficult to effectively pay lawyers to shoot their own feet.
That's how radiologists could analyze x-rays and such remotely: they were actual US-trained radiologists with licenses in US states who preferred to go home to India.
In most states, you have to go to an accredited law school in order to even sit for the bar, and all of such law schools are located in the US (and charge commensurately-priced tuition). I believe there are ways around this in medicine because you can go to medical school elsewhere and then sit for boards in the US. Even if you have a foreign law degree, you still have to get a 1-year LLM ($60k) in order to sit for the bar.
OTOH, after sept 11 my cousin (or, I think, his wife) decided they didn’t want to raise kids in NY so he moved back to Mumbai, but kept his job with a major ibank and salary, just working mostly remotely. I suspect though he’s an exception.
I've mentored aspiring entrepreneurs in the legal tech + services space for a number of years now and go out of my way to provide additional guidance (and some well intentioned ribbing) to those who pitch any sort of legal referral idea.
The concepts are (without fail) derivative and lack novelty, equivalent to pitching "Uber rideshare, but for [INDUSTRY]" without more to sophisticated VCs.
The problem is very real, but the margins are so thin that venture shouldn't be anywhere near it. Solutions should be non-profit and left to those heroes in the A2J space (whose funding should be significantly increased).
[1] https://twitter.com/justinkan/status/1191821878567915520
That's an injury that will keep you in bed, giving you a moment to really look if you're collectively as a company making the best effort on solving the right product/market fit. Which he clearly couldn't find, nor roll off some of the law-firm tools they built I guess? But the actual law firm of lawyers they got together is still running which is good, so it's not a total loss, plus some money back to investors.
It's difficult to get proper perspective sometimes being a founder, you get mired in details and fantasy growth possibilities, and startup runners constantly need re-committing themselves for the tough fight going forward. It takes balls to finally admit defeat, especially with a solid track record.
So ultimately I wouldn't be surprised if this gave an opportunity to step back and take some perspective, while he's stuck in bed in recovery mode.
It's a lot easier to experiment, learn, and iterate when the team is small. Gets very difficult when you're 100+ so quickly.
But the ability to pivot was obviously hamstrung. The team necessary for “plan b” didn’t have much overlap with “plan a.”
In this case, I think the headcount was a bug rather than a feature.
And
"My experience with this was also negative - seemed like a good idea but post-sale often took weeks, even months to get results" https://techcrunch.com/2020/03/03/atrium-shuts-down/
I think it turned out ok... time will tell I guess.
While I would have been happy to pay for this sort of concierge approach, I was also "declined" for being too small / not needing the specific services I suppose they were willing to provide. I imagine this kind of interaction with potential clients is one reason they got to this point.
The major work is still done by humans because of the liability, complexity and customization of every case. The next step requires a revolution in legal process, not just more software.
If there's really such a better execution to be had then Atrium was in the best place to find it. Yet it didnt.
1) It's shockingly rare to see use of project management tools, even simple stuff like Basecamp or Notion. When multiple people need to, say, review a draft of a filing in sequence, they're typically either keeping track of it in their heads or in a spreadsheet. "Excel is totally adequate for this" is wrong for a variety of important reasons.
2) Infrequent use of file indexing, even Windows' built in. I see firms using document management software running old versions of SQL with totally the wrong kinds of indexes and which usually take forever to find anything or crash in the process.
3) If they exist I haven't seen them in use, but a lot of teams would benefit from some middle ground between nothing and e-discovery software for searching across relatively large but not enormous collections of documents. A rudimentary on-prem or local CRUD app that could ingest a variety of shit, OCR it if necessary, dump it into Elastic, and let you FTS it would probably sell like hotcakes to small and mid-sized firms.
While the legal and tools companies were together, I'm sure the lawyers enjoyed the shiny tools. But after the "pivot", with the lawyers seemingly going it alone with some of the former clients, it seems likely to me that Atrium (the tool) just wasn't offering enough of a value proposition to make the tool company profitable.
Venture money needs massive growth. Selling a tool to lawyers will get an amount of money per month proportionate to the number of fee-earners. And from the sounds of things, their tools weren't exactly going to be irreplaceable. Anyone got any direct insight as to if the tools themselves were compelling but the pivot failed, or if the tools didn't deliver?
It wasn't quite what I thinking as it isn't really "our game startup failed; let's sell game tools" but still it counts.
This is different from either providing a service and successfuly using your own software to do so or providing a service and developing as well tools specifically for your customers.
Slack was suggested by another poster as a company that pulled this off, though I'm not sure the communication tool was considered crucial to the game business.
Not trying to attack -- just typing out the questions and consequences to think about.
It's hard to detect my own bias of course, but this seems like a really cynical move. I understand the need to improve margins but essentially trying the train our processes so we can fire you is off-putting and borderline offensive to workers. For positions closer to the furnace of automation and jobs that are generally dangerous that people don't take pride in it can be minimally destructive but law is not one of those professions.
It's preferable to get into businesses that don't screw over anyone, but if you're going to screw over someone maybe don't screw over smart highly paid professionals that are providing your core value proposition.
And the customers who want legal help for cheap(er)...just don't seem like the people who you make money off of.
Any customer concerned enough to know about this service probably wants more from it.
Anyone that would be a good customer, might never find them / isn't doing much in that area anyway...
I wonder when Justin lost his conviction? I don't believe you can lose conviction in 2 month.
One of my earlier startups had multiple acquisition offers which would have made all the shareholder employees quite happy, but since the offers wouldn't have moved the wealth needle of the already wealthy founders by an order of magnitude, they turned them all down.
Years later and $100m+ of VC money burned the place went bankrupt and I now have some very expensive toilet paper in a folder somewhere.
This is distressingly common, and not just in the valley.
I've seen numerous film deals fall apart because the producer didn't want to take a 10-30% discount on their standard fee (we're talking in the realm of a $150k discount on a personal take of 500k-1.5m on a deal), films that otherwise would've actually been made ended up falling apart because someone wanted to extract just that much more value.
Kinda like overplaying the market, really.
I'd rather pitch into the latter, which I suppose is sunk cost fallacy manifest.
> what is "blood equity"?
Even failing startups, with little cash and poor market penetration are often worth something to somebody. It could be a small client list, the team, patents, or just an attractive domain name. Most founders leading failing startups will try to sell to suitors for pennies on the dollar invested. Sometimes they get a small cash bump (or in WeWork's case, a small $1B), but it also means they can say they were acquired rather than simply failed. And they're right, building a company and selling it for $1.00 is far more difficult than not selling it at all.
Investors don't always like this because they recognize the loss at acquisition, which hurts their published returns. But more often, they don't care. They know the odds before-hand and prepare for events like this.
The people who lose the hardest are the employees with just stock-options or small equity stakes. Everyone's stock will be worthless, but the founder can potentially walk-away with a decent "consulting" contract. The best a standard employee can hope for is being hired by the acquiring company and that their culture doesn't suck.
From the employees perspective, the bright side is that they are often the first people to figure out that a company is doomed (before investors, customers, or the press). They have the advantage of time when it comes to planning their next move.
part of the “challenge” of being a famous founder is you can raise pre product.
$75M seems insane for a company like that. Burn couldn’t have been that high, mostly rent and salaries, no heavy computation, etc.
Weirdly, the investors are happier with that than 5 smaller series A's of money-making / real-IP companies with existing growth engines. Easier to bet on nothing vs improving.
I do find the flameout odd, and not isolated to Atrium. Why flameout vs spinoff a small business for folks who wanted to continue? Same difference to the investors, but meaningful to the lives of the believers. Worst case, just OSS for them..
Thinking more: A good comparison is Carta, also in legal, whose funding was more correlated to revenue and growth.
In the case of Atrium, I'm sure that the investors would have had plenty of board-level conversations about the potential trajectory of the company. The decision to shut down and return funds was likely endorsed by the investors, if not proposed by them, who would doubtless remain extremely keen to invest in whatever Justin comes up with next.
[1] https://techcrunch.com/2014/08/05/justin-tv-shuts-down-to-le...
But hindsight is always 20/20
tools and software for patent law are abundant and low value. you can't just say "tools and software for [INDUSTRY]" and assume a business model exists there. you need some insight into how firms operate.
> aggregating data and knowhow
law firms are not going to give a startup access to client data. especially in any format that could be aggregated. 0% chance. confidentiality, privilege, and patentability, are essential. it goes against our business model and professional obligations.
> and then pivoting to an “automated” law firm as a service.
the portion that could be automated is 10% of the value. experience, both with the practice area and the client's industry, are 90% of the value.
think of stackoverflow, yes an applicable code snippet exists on there, but understanding how it works, what its going to do, and the larger implications is the valuable bit. you can't automate that.
If the legal services arm was able to license the name and technology in a permitted way (like how a law firm can of course get their website and IT done by experts), it would seem to present an experiment at whether such a structure could act as a conduit to channel fees back into an investment vehicle.
Most startups will want to build a relationship with an actual lawyer, and having something like a LegalZoom for startups didn’t make sense for most.
Legal industry already has plenty of good tooling though. There's not much more to do here outside of automating the lawyers, and that won't happen.
Because at the end of the war or a death of a startup, everyone is a field marshal analysing the failure afterwards.
>"It will return some of its $75.5 million in funding to investors, including Series B lead Andreessen Horowitz."
Does this simply mean it will return all funding after paying out remaining invoices and salary obligations? Or is there some significance to the phrase "some of" here?
So A16Z isn’t getting back anything close to $75m. Just “some.”
Part of this seems like courtesy among the big boys who want to play again. I doubt Atrium wants to tick off A16z by making dubious claims to the remaining cash.
Unfortunately, if you even think about your attorney, you'll be charged for it but the peace of mind is worth it.
Having a lawyer you trust, is timely to questions, and works fast is just a layer of stability you need so you can focus on the important parts of a startup. Legalzoom is garbage in this regard.
The product makes you feel like it’s a one click thing but you need to check in with a lawyer to not forget anything. Most ppl who used Atlas have told me the same. I do recommend Atlas since you will save some money and you get some nice fringe benefits.
Cost also doesn't guarantee quality. You're paying for resources, specialization, general counsel that can delegate, and insurance against mistakes. Useful but only when you really need it.
I’ve seen agreements that were 99% good, but a problem in the other 1% made the whole thing worthless.
RocketLawyer is even closer to that: https://www.rocketlawyer.com/legal-documents-forms.rl/
$300 might cover a simple contract review, maybe by volume you can cut that in half - so if you are talking 1-2 "tickets" a month, maybe you are covered. Sounds risky to me in a back-of-envelope way.
Of course, a bunch of people are trying variations of this idea.
You can find online legal guesswork and forms for free. $300/mo is too expensive for that.
But getting someone with experience to stand behind it - to risk their license, insurance, and future income — is north of $300/hr even in suburbs and flyover country.
The main selling point was that it would re-create state forms on the web and allow an attorney to fill them out, print them, and manage a case.
These forms would change many times a year and they were written using horrible PHP libraries that required every piece of the form to be drawn using coordinates. If one thing changed on the page (which happened often), all coordinates would shift and have to be re-written.
Aside from this, the owner had the first version written by developers overseas a few years earlier and I was hired as part of the 2.0 team that consisted of more inexperienced overseas developers and resulted in even more of a mess than version 1.
Some gems from the owner:
She once put on her Linkedin profile that one of her accomplishments was the ability to hire developers on a volunteer basis or well below marker value. It was taken down within a day.
I was finally let go and was told straight to my face that she was going to hire someone from India for 1/3 what she was paying me.