When I saw the post by Bob Ambrogi, I was, frankly, stunned. Twelve million dollars? How could that be?

The legal research service Casetext is announcing today that it has closed on a $12 million Series B funding round. Excluding e-discovery companies, it is one of the largest investments in a legal technology startup ever.

That’s a lot of money even for stupid money, and yes, Bobby didn’t make a typo in the name. It was Casetext. Yes, that Casetext.  The legaltech start-up’s model was to offer legal research with court opinions annotated by crowd-sourced analysis and commentary. In other words, any moron could go on Casetext, “explain” what a statute or decision meant, and any other moron would then read it and believe he now understood law. It was absurd. Dangerously absurd.

In the beginning, I let the founders, Jake and Pablo, scrape SJ content mostly because I am all in favor of anyone bent on breaking the duopoly of Lexis and Westlaw.  But that only want so far.

But because I hated the duopoly of West and Lexis, and want to help anyone or anything the screws with their business model of selling us back our judges’ decisions, already bought and paid for by the public, I told Pablo and Jake that it was cool, they could glom my feed (Edit: After Jack Heller read this post, he pulled down the SJ content from Casetext).

Then I found out that they were running a comment section there, even as I was at SJ. I told them to pull the plug on that crap immediately. I got emails of the comments (which were beyond idiotic) and had no intention of going to yet another platform to tell stupid people to get their heads out of their butts.  They complied.

It wasn’t just that Casetext was an idiotic concept, or that their second foray (“Legalpad,” a pseudoblog) into having something to sell wasn’t similarly stupid. It was that Jake was one of those snot-nosed children who woke up one morning thinking he mattered.

I received an email from Jake Heller, CEO of Casetext, which included this gem:

If you’d be interested in chatting about it sometime, I’d be happy to share my thoughts.

To which I delightfully responded:

Why?  We’ve talked. There is nothing I’ve said in the post that I haven’t already told you. But why in the world would I be interested in chatting about it with you? How utterly fucking arrogant of you to suggest that you are “happy to share your thoughts.”

It’s you who wants to talk with me. And you’re more than welcome to my advice and counsel, when you pay me for my time. But to suggest that I might want to chat you? Are you really that clueless?

Now some youthful lawtrepreneurs are flaming, self-important narcissists by accident. They’re well-intended, but dopey enough to believe that they’re very important people.

Heller, on the other hand, isn’t such a doofus. He’s a smart kid, and when he tries to flip it to suggest that he’s happy to do me a favor by giving me his spiel, it’s deliberate.

Not that pathological narcissism isn’t transmitted by tummy rubs, but this was more arrogance than I could take. Thereupon, I crossed Casetext off my list of things worth another second of my time.

But when I saw Bob’s post, and after verifying that this was the same Casetext, I had to find out how this was possible. Could I have been so utterly wrong about the stupidity of the business model?  Bob explained that while I wasn’t paying attention, Jake apparently stumbled into doing something right.

The legal research service Casetext is unveiling a new service today that automatically finds cases that are relevant to legal memoranda and briefs. With this unveiling, Casetext, which has been free to use ever since its 2013 launch, is also preparing to roll out its first paid subscription tiers for premium services, while keeping basic access free.

The new research tool being unveiled in a limited rollout today is called CARA, short for Case Analysis Research Assistant. What it does is find cases that are relevant to a legal document but not cited in the document. Upload a brief, memorandum or any other document that contains legal text, and CARA analyzes it and generates a list of relevant cases that are not mentioned in the document.

The new model is AI-assisted legal research, with panels appearing to provide blurbs in subsequent mentions of the case. Now this is interesting. Not only does it create a monetizable product (priced at $119 per month per individual lawyer), but it actually offers something someone might want.

But Jake gonna Jake.

From the start, Casetext has focused on improving legal research. As a litigator, Casetext founder Jake Heller experienced firsthand the frustrations of expensive, outdated, and inefficient legal research tools. And as an engineer, Jake understood that applying the best technology – data science, machine learning, natural language processing – could make things much better.

Notice what’s not in there? Anything about its original failed model. But hey, why screw up a viable concept, right?

Bobby tells me that they haven’t entirely forsaken the ill-conceived crowdsource the law idea, and it’s still in there. Why is unclear, unless it’s as simple as Jake’s arrogant inability to recognize that it sucked from the start and admit his baby was born ugly and only got better looking later.

But the fact appears to be that Casetext, for all its flaws, managed to pivot into something that may well work, and may, in fact, actually succeed. For now, it’s got $12 million to work with, and that’s not chump change. Whether the baby lawtrepreneurs will grow up fast enough to let go of their bad ideas and make the good one work has yet to be seen.

And Jake, even though you seem to be on to something here, don’t mistake kismet for genius. You’re not doing us a favor. You just want lawyers to give you money, like every other arrogant little boy in legaltech start-up land. Casetext might have successfully pivoted. Have you?