Here are some of the lessons I learned from Sabaio.com and Legalcomplex.com.

What Is Sabaio? In two words: Private Intelligence. Sabaio is an AI platform that leverages open-source software such as Ollama and vLLM, running in Docker or Kubernetes and compatible with Windows, Linux, and macOS. It is a flexible framework that allows enterprises to deliver basic AI capabilities like proofreading, summarizing, and building RAG workflows. Sabaio also supported advanced legal tasks, such as generating synthetic legal data, legal reasoning, contract workflows, and compliance research. Sabaio was deployed at a 400-person European law firm with single sign-on.

What Made Sabaio Special? Sabaio runs completely on-premise and does not require internet access. It processes all data locally and does not track or share any user data, making it ideal for governments and other sensitive sectors such as healthcare, banking, universities, and legal.

When the DeepSeek app rocketed to the No.1 spot in the US iPhone App Store, Americans were quick to push for legislation banning its cloud version. Using cloud AI is not necessarily unsafe, but the key question is: Which country is hosting your cloud AI?

One would assume a Western country is safer. Yet, the UK has issued a secret order requiring Apple to install backdoors in iPhones and iCloud – effectively giving the government access to all iPhone users’ data globally, (as reported by The Washington Post). Essentially, any government can compel a cloud provider to provide access to all the data it processes. AI is the perfect choke point to tap into human activity.

Why Did Sabaio Die? Despite being open source and solving an obvious need, the economics didn’t work out. Running AI locally requires specific expertise, sufficient electricity, and expensive computing power. At a minimum, Sabaio required a single NVIDIA H100 GPU to serve 100 users with a 70B parameter large language model. Scaling hardware to accommodate more users required specialized infrastructure, known only to a small group of experts. Beyond talent and hardware, there are also cooling and energy costs.

All this requires significant capital. Running on-prem AI means recreating the same data center infrastructure that Microsoft, Amazon, Meta, and Google are frantically building. These four companies alone are projected to invest $320 billion in AI in 2025, as reported by The Wall Street Journal. Imagine how much a single enterprise must allocate to support its employees.

Why Not Raise Venture Capital? In my humble opinion, that would just add fuel to the fire. The customer acquisition and onboarding costs for Sabaio would force me onto a fundraising treadmill with no clear exit in sight. Additionally, the slow onboarding process of on-prem AI doesn’t align with the ‘grow fast’ venture capital mentality.

VCs are also more comfortable funding Software-as-a-Service (SaaS) models that charge licensing fees. An open-source business model, which doesn’t rely on licensing fees, instead requires a large user base to convert. The private intelligence market is relatively small and firmly locked into the Microsoft ecosystem.

Even though the premise of distributed, decentralized, open-source AI is perfect, there is no sustainable financial model for it. While AI costs will decrease as adoption increases, future cheap AI doesn’t help offset today’s expensive infrastructure costs. Everyone will eventually need to recover their initial investments – but when will that point arrive?

What’s Next for Raymond? I will wind down Sabaio and Legalcomplex. The last Legalcomplex analysis will be the Q1 2025 report.

This 7-year rollercoaster has been an incredible journey. I’ve had the opportunity to travel, meet amazing people who loved Legalcomplex, and develop skills in financial analysis, identity access management, LLM fine-tuning, and LLM orchestration.

I’m a relentlessly resourceful, creative introvert, and I’m honestly excited to start a new chapter in my life and embrace a new challenge.

Originally posted by Artificial Lawyer on Feb 11, 2025
Added links and minor edits

It has never been cheaper to get AI. AI funding is up, and so is revenue, but what if costs keeps increasing, will the bubble pop?

What is a Bubble?

Historically, bubbles have been characterized by exponential growth in investment with lagging revenue growth. In plain English: a lot of money came in, and almost nothing came out. This happened during the dot-com bubble of the early 2000s. We expected the internet to fuel commerce. Eventually, the internet did give us Amazon and Alibaba, but the first winners were Google and Facebook. So search, and social, came before commerce. Yet, the only winner during the dot-com implosion was a hardware company called Cisco. Cisco stock price soared to 3,800% from 1995 to its peak in March 2000, making it the most valuable company globally at that time. This also happened to NVidia, another hardware company. What are the odds of history repeating itself?

Dot-com was not the only bubble. We also had Covid-19, which brought us low interest rates and shelter-in-place behaviors. This drove investments in companies sky-high. Once life returned to normal, all valuations came crashing back to Earth. No other company exemplifies this more than Docusign. Due to its rapid rise during the pandemic, Docusign was included in the Nasdaq-100 in June 2020. However, its listing lasted two years, and it got kicked out by December 2022. Yet, Docusign is one of our most reliable canaries.

So the question is, are we in an AI bubble? Let’s calculate some numbers and look at the graphs.

What is Venture Capital telling us?

October 2024, we registered a record five fundraise announcements in a single day. This year, we witnessed a billion raised by legal startups in a single day, another record. We made a video chronicling all historic legal fundraises over the years starting in 1984 with Adobe. The company that created PDF: the only way legal could replace paper to securely share contracts. Bringing legal from stone tablets into the digital age. Sorry, we digress. History is the past, and we’re now in the present. So how does funding look in 2024?

Glad you asked, since we have a few more records to announce. The graph above shows us the total raised by legal companies in the past six years in each quarter. You’ll spot that (a) Q3 2024 is the highest ever Q3 total and (b) ranks third overall after Q1 2021 and Q1 2022. Yet, as we reported in reaction to investment bankers, the total number of companies getting funding is dropping. Translation: more money went to fewer companies. This brings us to another new record: at no time in recorded history have legal companies raised this much money on average. Investors clearly feel confident that they are picking winners. Same winners as in the Dot-com and Covid-19?

We mentioned bankers, which brings us to another metric most overlook: Debt. This year, we are on track to set a new record in bankers offering credit to legal ventures. Same as funding, fewer companies received more in loans than at any time in history. That is $4.1 billion in credit which is up 130% over last year. Even with interest rates on loans at an all-time high, bankers believe in legacy companies to survive. If you are interested in our take, read our rundown of them all here: Data, Distribution & Deals. Spoiler: we disagree. We could be wrong, and here’s why.

What is AI revenue telling us?

The first line in this piece ends with:”…lagging revenue growth“. The revenue generated from AI applications in the legal field is the critical metric. A simple proxy to calculate growth is OpenAI revenue. It has been the fastest company in history to a billion in annual revenue. That’s why it is now arguably the highest-valued private company in history. However, it is also the most capital-intensive company ever and projected to be profitable by 2029. Meanwhile, their expenses will triple to $14 billion by 2026. But who is receiving $14 billion from OpenAI?

In hindsight, history had already answered this question, yet it took us two previous posts to unravel. The answer: producers of general processing units (GPU) and electricity. Now, a scientific breakthrough in either will trigger the biggest wealth destruction we will ever witness. That is the Black Swan scenario. Meanwhile, all AI companies, including legal, are funneling money towards GPUs and energy.  Economists at Goldman Sachs calculated about 4–16 months of runway to keep spending on AI. If earnings do not keep up, how can one keep paying expenses?

Here’s a thought experiment: there is a nuclear power plant that does not produce power because it is closed due to the biggest nuclear disaster in the Americas. Here’s a pop quiz:

• Why did a software company (Microsoft) buy this nuclear power plant (Three-mile Island)?
• Maybe because Microsoft sees energy as its new source of revenue?
• Perhaps software revenue will decline over time?
• Is it that AI, will eventually write all software?
• Is Software-as-a-Service (SaaS) dead?

What revenue is telling us, is to look at the costs. With AI, the production costs are concentrated. If we could only decentralize and distribute the AI costs. Where did we hear this before? Oh, I remember: Why the World Needs Decentralized Legal AI.

Author note: this may be my last post. AI will flood the airwaves, so posting will become futile. I wrote these posts as a diary, to remember.

Most companies selling AI will not be profitable anytime soon. Nevertheless, we have numbers to measure how valuable AI companies are.

TLDR: In Data, Distribution & Deals, we looked at the value of public companies operating in the legal sector. Now, we’ll do the same for companies in the private sector.

Validate

Brace yourselves, for this one is stacked with numbers. Starting with two lists: AI companies and Legal companies. The listed examples underscore the current market conditions:

• Inflection AI: Raised $1.5 Billion in funding, but reported negligible revenue. Microsoft agreed to pay Inflection AI $650 Million to acquire the tech and talent;
• Stability AI: Their expenses exceed its revenue, with spending estimated at $8 Million per month. They already raised $256 Million and are discussing a potential sale to private equity;
• Humane AI: They are looking to sell the company barely a month after the AI Pin release. Humane raised $230 Million from prominent investors like OpenAI’s Sam Altman.

What is happening in Legal? The chart above shows what we have for this year.

• Septeo: Private equity firm Hg reportedly wants to sell this $4 Billion Legal Tech company;
• DocuSign: acquisition talks to private equity were reportedly halted around February 2024;
• Eigen Technology: One of the most well-funded Legal AI startups, with $59.4 Million in total funding. They were acquired by Sirion Labs for an undisclosed amount;
• Henchman: Belgium contract drafting tool was acquired for $160 Million by LexisNexis. Henchman was founded in 2020 and has raised $11.4 Million in total.
• vLex: acquisitions talks with Harvey AI collapse for undisclosed reasons. Before the collapse, Harvey was set to raise $600 Million but now projected to raise $100 Million.

Legal exits have been on a downward slope since the beginning of 2021 as illustrated in the chart below. Statistically, only 19% of law acquisitions reveal their price as observed in the 2022 analysis with Law360. The amounts above offer a glimpse into the market value of companies.

Valuable

Henchman, and CaseText acquisitions should be celebrated for their transparency. Respectively, the Henchman was 14x return, and Casetext achieved a 10x return on invested capital in those companies. The return is calculated by dividing the purchase price by the total amount raised. For Henchman that is $160 Million divided by $11.6 Million equals 13.8x to be exact. It is not NVidia territory, but is it a good return? Let’s see.

Valuations are visible when an acquisition price or share price is offered to the public. Those valuations are calculated based on existing revenue. Case in point: Google announced interest in acquiring Wiz for a reportedly $23 Billion. Wiz’s annual recurring revenue (ARR) is reported to be around $350-$396 million. Using the higher estimate of $396 million and the proposed acquisition price of $23 billion, we can calculate the revenue multiple ≈ 67x. However, Wiz has raised $1.9 Billion since 2020. This would offer their investors a 12x return on investment (ROI) within four years.

For some context, “The Best Acquisitions of All Time“, made by Acquired, offer ROI guidance for multiples. Spoiler: the winner is 153x return of investment (Instagram by Facebook). So how does legal stack up? Check out the chart below. Better yet, the median ROI for the tech sector is 5.03x, measured across 515 deals. For the companies in legal, the ROI is 8.47x, calculated from 58 deals.

Voodoo

This brings us to the most mysterious process in venture fundraising: pre- and post-money valuation. As stated at the start, the Harvey-vLex deal fell through. As reported, Harvey would have been valued at around $2 Billion if it had acquired vLex. Without vLex, the value is $1.5 Billion. Simple math reveals the valuation of vLex to be estimated at $500 Million, as pointed out by a commenter. This is less than CaseText and perhaps the reason their private equity owners walked away. Harvey has raised $106 Million in total. If Harvey were to be acquired for $1.5 Billion today, the ROI would be equal to Henchman (14x). Hence, this is a reasonable multiple for Harvey but what about vLex?

To answer this question, we compared the size of CaseText to vLex based on a rough headcount. At the time of acquisition, CaseText was half the size that vLex is now. So vLex valuation would be closer to $1.3 Billion. You can imagine the powerful private equity executives leaving the negotiations laughing. In the snippet below, you’ll read the reason Harvey has not raised is a hesitancy towards AI investing. We posted a Goldman Sachs podcast that elaborated on the AI bubble scenario. In our previous analysis, we explored the value that AI will deliver. Short answer: productivity, but no profits yet…except for one company in legal.

The future of AI is currently cloudy with a chance of rain. According to Goldman Sachs, we have between now and the end of 2025 to figure out a path to AI profitability. Meanwhile, we will dream of all the possible products that will push legal into the future.

Where is all the AI money today? Nvidia. And tomorrow? We made a new AI Value Stack to uncover new business models.

TL;DR: AI is shifting volume and value in the tech market, altering the profitability of existing business models. The first to feel the impact are accounting and law. So these industries have to pivot to new models.

Value

Let’s illustrate some business model pivots with two stories:

1 PwC became an OpenAI official reseller. Foundational AI companies like OpenAI can not raise more funding to train and operate Large Language Models (LLM). It is simply too expensive for anyone to finance. Therefore, OpenAI will have to fund AI from revenues. The shortest path to big checks is by attracting large enterprises as customers for AI. The softest spot to target is legal, since legal has weakened competitors. It is the area that avoids SAP, IBM, Oracle, Microsoft, and Google as competitors. That is why the first video promo from OpenAI showcased a contract review GTPs.

Remember, OpenAI’s goal is to build AGI for consumers, but they need corporate money to fund it. Consultancy provides an easy entrance into corporations and governments, as they’re already embedded. Traditionally, the business model for consultancy relied on throwing bodies at a problem. Now they are pivoting to throwing bots. Consultancy firms will experience declines in demand when customers hire more AI. Note: the Wall Street Journal (WSJ) reported struggles at Mckinsey, arguably the world’s biggest and baddest consultancy firm.

2 Gartner predicted the legal tech market size will double by 2027 from $22.3 Billion to reach $50 Billion. Luckily, we kept track of market size predictions. We calculated market size in 2023 at around $40 Billion, and grew to that size over a century. Market size estimations have one drawback: they rely on existing business models and markets. They do not consider business model pivots or new markets. Now, will AI spur growth by increasing the value of legal tech companies? Actually, the contrary. Please, keep reading.

The value of a company is determined by velocity of a market and the volume of products. In some cases, a market may not even exist. The AI chip market did not exist a year ago. As of June 3rd, 2024, Nvidia’s market cap is $2.7 Trillion (with a T). Since Nvidia owns nearly 100% of the new AI chip market, $2.7 Trillion is the market size. The demand for AI is here to stay, but the lead that Nvidia has, will not last forever. So how can one better project the shape of any market?

While presenting in Estonia last year, we mentioned Nvidia as one of the emerging AI players. Using a CAT scan on recent investments, we showed the audience the saturated and emerging markets. Basically, the CAT calculations help us follow the smart money in Spark Max.

Volume

This brings us to the other part of the equation: volume. The more frequently a product is used, the more valuable it becomes. Search, social media, and smartphones are examples in the consumer markets. E-signatures, cookie consent, and filing taxes are examples in the legal market. The increased volume of transactions in these products created new markets. Now, imagine that almost all the interactions will go through a single interface. That AI will intercept more questions and answer them without the help of other services. Subsequently, the volume of use and traffic to other products declines. What happens to those markets?

Here’s what happens: Stack Overflow answers a huge volume of technical questions. Traffic to Stack Overflow plummeted when visitors adopted various paid and free AI coding copilots. Stack Overflow’s business model is primarily ads. While they initially were against AI, they struck deals with Google and OpenAI. This created an interesting business dilemma: if fresh answers are declining, will the value of Stack Overflow also decline? Well, Reddit is now a public company. As a similar crowdsourced platform, they will provide insights into this new business model experiment.

Some may argue that content cannibalism only impacts crowdsourced platforms. Highly specialized, proprietary data may actually be worth selling to AI providers. I was a firm believer until I heard why this is wrong. Around the 50-minute mark on YouTube, a16z’s Ben & Marc expertly explain that selling proprietary data to AI is suicide. Selling legal data is especially problematic, since legislation and court data, are constitutionally already free. As noted, Copyright will not shield owners, and we may even get new laws to require legal data to be added to AI. If that happens, we’ve entered universe two. So what business models will work in the era of AI?

Visualize

To answer this question, we’ve designed the AI Value Stack. The AI Value Stack is a framework that breaks down the various layers required to build, operate, and sell AI solutions. This helps us to visualize the production costs and to estimate profit margins. Each new AI advancement endangers current products on the market. Especially, when companies rely on a single supplier. So this framework will also future-proof AI products by eliminating supply chain risks.

Let’s take the Energy layer to illustrate: a popular AI business model is the computer coding assistant. Microsoft GitHub Copilot charges $10 but actually costs Microsoft $80 to produce, according to the WSJ. One reason is that a single AI query consumes 10x more electricity than normal cloud operations. However, if it’s only a coding assistant you need, you can run a copilot on your laptop. Better yet, running it locally is not only more energy efficient, it’s also free. This YouTube channel drops new free alternatives every week.

Free AI is not yet a reality in the high-end financial and legal markets. Legal AI pricing is rumored to be higher than even the premium legal tech products. One reason for the extreme pricing is to offset future revenue declines and maintain margins for consultancy and law. The other reason is the cut that AI providers take. Essentially, Legal AI pricing relies on the current business models and is therefore flawed.

Conclusion

The AI money is running out. Even Microsoft will lay off 1500 as they announced a $100 Billion data center. A data center that will handle all AI queries and needs around 5 Gigawatt of power. That is enough electricity to power 4 million households or the entire city of Berlin, Germany. Such a data center would be hard to host anywhere in Europe. Hence, EU AI queries will not be processed on EU soil. Unless we use smaller models that run everywhere.

But, how do we make money with AI? Well, clearly not by charging for wrappers, like GitHub Copilot. Selling data also would not last, as seen with Stack Overflow and a16z. Similarly, the Nvidia lead is temporary while competitors ramp up production. We’ll need new business models that consider the dynamics of volume, velocity, and value of markets. In simple terms: a clear breakdown of costs per layer, the competition per market, and momentum.

In closing, I’d like to leave you with this tweet posted on October 14, 2013, by Box founder, Aaron Levy.

Incumbents are rarely disrupted by new technologies they can’t catch up to, but instead by new business models they can’t match.

Aaron Levie

We now ask AI to answer complex questions. Eventually, AI will also take actions on our behalf. At that point, will we lose free will?

Borg

My eldest just discovered Black Mirror, and she asked for my favorite episode. It’s impossible to pick favorites, but I eventually settled on season 3, episode 3: Shut Up and Dance. No spoilers but the overarching theme of Black Mirror is the impact of technology on society. Especially the pressure tech exerts on our power to decide. A power which is derived from access to our information.

The main reason TikTok is banned in the USA is the fear of Influence by China. The irony is that this move is unconstitutional since the decision to ban is actually also authoritarian. There is legal precedence for reciprocity. Meaning, China bans American apps all the time. However, the speed at which these decisions were made, invoke unease. Especially with a lack of evidence and transparency on all sides.

To summarize: Technology with our information, can exert control over us. That technology is hard to trust when control is concentrated in a few places.

AGI

Sam Altman, the CEO of OpenAI, defines AGI (Artificial General Intelligence) as an AI system that is generally smarter than humans and can perform any intellectual task that a median human can do. Whether you agree with this definition or not, such a system would require immense energy, data and compute resources. Not only all the world’s knowledge but also all our personal information. Who we are, who our friends and colleagues are, where we work, and how we feel. If we are hurting and ask AI to heal, it has to know our symptoms. And it will need to remember more about us, to get better at serving us.

AGI will need more data on us because we’ll rely more on AGI for our decisions. And one day, AGI will also take actions on our behalf. In case you missed this: The Rabbit R1 is a physical device that can order a meal from DoorDash. Now, we should not expect these devices to be flawless any time soon. However, consider how we could ever live without them after they do become telepathic. He rabbit, how did you know I was hungry and craved some taco’s?  Rabbit: you have habits and there is a pattern. You are predictable.

The ability of AGI to anticipate our every need is scarier than the capability to wipe us out. Now, the first to give us AGI will receive the most power to shape our lives.

Remember

As humans, our natural drive has been to automate as much as possible. AGI is the tool that brings unimaginable efficiency. This will change business models and our entire economy. Society will shift from Google/Meta attention economy towards an algorithm economy. Like TikTok, those who have data on our habits will have the best chance to break them. Similar to the ban, it is technically and psychologically hard to untangle the dependencies. Unless we decide early what would be the best design for these systems to influence us.

The rise of Blockchain coincided with the Occupy Wall Street movement. The solution that made blockchain technology so attractive was its decentralized architecture. This meant a distribution of power. In short, a few unelected could not mess up the lives of the many. This also attracted the legal industry, which invented a concept of separation of power dating all the way back to Greece’s Aristotle. Only lawyers feel a twinkle when they hear the name: Montesquieu. Ultimately, a long line of legal scholars wrote the code that makes society run safely.

Ponder: What would be a healthy and safe design for AGI to rule us all? Well, if you’ve had any legal ed, it should be obvious.

While everyone sprinkles spicy AI sauce on their product, some realize that their customers’ appetites are changing. So, should you pivot?

TL;DR: AI presents current products with this warm blanket of extra usage. However, more AI usage on top will impact the bottom line. Tokens look cheap now, but they are heavily subsidized. Hence, the question: where is a product’s value if it needs AI?

Burn 🔥

Why are we talking pivots? We never did before, plus now is the best time. Between 2020 and 2022, about 953 companies raised funds in 1546 rounds, taking advantage of the low cost of capital. Yes, some raised multiple rounds due to low-interest rates. In 2023, the music stopped. However, those companies have to keep dancing. Now, focus on the little flame, where it says ‘Burn‘ in the dashboard above. The 120 number represents the median number of days between rounds during that stretch across all fundraising activity.

We’re proud of our Burn metric in our dashboard. It estimates when a company will need more capital after fundraising. Burn is influenced by the type of business and geography, and is not equal for any two companies. Case in point: a legal tech company in Brazil becomes profitable faster than its counterpart in the USA.

Ultimately, Burn is the most probable answer to performance for any segment, area, and geography. Probably, a large language model will come up with a better indicator for performance. Perhaps that will be a sign to pivot.

In general, we’ll keep track of fundraises especially of Debt Financing. Since Debt is a strong indicator of economic health over time. Debt also fuels the greatest power in our industry. Debt not only causes us to lose friends, but it may also cause friends to lose jobs.

Value 💰

We’re chatting off-air about me being invited as an expert on a radio show. I said: “nobody can realistically call themselves an expert in anything anymore”. AI has now surpassed our knowledge capacity in almost everything. I’m only slightly more reliable at processing speech in real-time. Many aspects of my value as a productive human are debatable compared to AI. A fact that some consultancy firms may already be experiencing. I made my peace and started focusing on what would be worthwhile. Part of this journey was looking at the most valuable legal business models in Data, Distribution & Deals.

Our analysis concluded that legal data will be completely absorbed by LLM’s and copyright can’t stop it. Distributing anything with AI looks cheap now because access to AI is heavily subsidized. With interest rates still high, deals are drying up. Currently, litigation is booming but mostly for Elon and OpenAI lawyers. Remember, ‘booming’ in legal is impacted by geography as well as in which season we are. More on where money is made in legal? Check the Capital & Conflicts analysis.

In short: At the moment, value in legal is fuzzy. Value is much clearer in segments like Governance, Risk, and Compliance (GRC) also known as Reg Tech. Hence, law firms building chatbots instead of buying legal tech. In case you missed our earlier hint: Legalcomplex will also pivot to a different model. Generating any deep insight is now firmly in the hands of machines. The future value lies in helping machines bring those insights closer to us.

Pivot 🌱

A pivot is a beauty and a beast. We can start fresh and shape products of the future. That means we have to start over and relive the pain of the past. On the one hand, we are not in a panic about allocating $8 Billion to rescue revenue. On the other hand, we don’t have that luxury either. We can not just tap $600 Million in credit to buy time. What does this all mean? Everyone is putting their chips on the table. With AI, everyone has a seat at the same table. The question is: Where will we place our bets?

We assist enterprises in finding opportunities by calculating investments. We use the Burn and Growth metrics to gauge each segment, area, and geography. There are roughly 1151 companies globally in Legal that need cash and are open to talk.

The image above shows us how many companies want to have a conversation each quarter. The number in each column represents when they burn out of capital and need to raise again. The quarter represents the time, we most likely can expect a call from them.

Call 📞

So we’re meeting with a few of the 1151. Helping them raise more capital or exit fast. During these talks, the topic of a pivot has come up. A pivot helps with pitching your product to investors and acquires. The reason is that a new direction would increase the value of a company. So how do we determine value?

The value of companies in legal relies on that other holy trinity: Targets, Talent & Technology. Presently, companies with income from customers (Targets), have more time to assess a new direction. Those with just tech and talent may need to pivot sooner. And steer away from burn faster.

To conclude: Eventually, everyone in legal has to pivot to value. The best time to pivot is now. Where too is up to you.

Sports need rules and referees. So does society. Interestingly, technology gave us more referees. So will we get more lawyers after layoffs?

TLDR: Yes, more legal professionals doing more legal work, but doing it differently. GenAI will displace traditional legal roles and tasks and ‘generate’ new ones. But first, we’ll need a rule change.

Queen’s Gambit

Here’s how we’ll structure this one: First, look at what lawyers do to understand why they become redundant. Then show what happened in other industries after tech came in. Finally, let’s explore the emerging roles impacting the legal field. Keep an open mind, and don’t let fear cloud your judgment.

Four years ago, we examined what lawyers do and why they will be displaced by GPT. Here’s a quick recap: a lawyer finds, drafts, interrogates, and explains legal data to a layperson. That is exactly the skills we saw GenAI master. A GPT can only perform these tasks accurately if it has access to correct legal data. To gain access, we have to change some laws. To improve accuracy, we should acknowledge our biases in legal data. Now, we are going through the displacement phase.

Understandably, it’s difficult to accept our skills becoming obsolete. Imagine how Gary Kasparov must have felt when he lost to Deep Blue. Interestingly, chess has become more popular than ever since then. In 2013, we wrote about Freestyle Chess and the symbioses of humans and machines. More recently, the pandemic hit series The Queen’s Gambit caused a resurgence in interest in chess. Ultimately, games are the best simulation of society, and we can use them to envision the future.

Top Corner

The Africa Cup of Nations (ACN) hosted 52 football matches, with only one notable VAR controversy. This was an amazing outcome for a tournament that has a rich history of horrible officiating. VAR is the acronym for Video Assistant Referee and is our favorite analogy for legal. This analogy is useful to reveal the flaws when using technology to decide right from wrong. And now we’ll use the same analogy to demonstrate how technology can create new jobs.

Originally, football only required one referee. Now we have up to twelve (12) officials, plus a fully automated referee managing major matches. Let’s list them here to illustrate this point:

• Main Referee (1)
• Assistant Referees (2)
• Fourth Official (1)
• VAR – Video Assistant Referee (1)
• AVAR Assistant Video Assistant Referee (4)
• Offside VAR – OVAR (1)
• Reserve Assistant Referee (1)
• VAR Technician (1)
• Goal-Line technology GLT (1)

That is eight (8) new human positions and eleven jobs created by technology. We haven’t calculated how many people work on developing and setting up VAR. Fun fact: GLT, is the only automated official, and also the only official with a spotless record. Makes you think, doesn’t it? So, why has technology led to more arbiters in sports?

Bending Rules

On the streets, everyone can call a foul, and no one cares. A Champions League final attracts 450 million viewers. For context, this dwarfs the 123 million viewers of the last Superbowl. That is a lot of angry customers when bad calls mess up a match. Contrary to fake football, we watch the game and not the TV ads or Taylor Swift. To avoid controversy, sports implemented rule changes and allowed more technology into a high-stake emotional experience. Unfortunately, in law, we aren’t ready to make those changes.

There are valid reasons why the law has not gone fully automated. In ALI, we discussed some academic observations and technical limitations. Notably, neither Perplexity AI, Gemini, nor ChatGPT with Webpilot were able to provide the 52-match number at the ACN. However, once we’ve fixed those bugs, and changed the rules, we’ll be watching an entirely new ballgame. In a world divided, can we agree to change the rules?

Let’s pick our favorite subject: Taxes. Why, and how much, has roughly remained the same for as old as civilization. To clarify “how much”: the poor always pay more than the rich. Even though the world is in constant conflict on everything, we did agree on one thing: a global minimum tax for companies. Imagine: 138 countries agreed they want this. This global tax treaty is a win for the legal profession. It shows that with our skills we can achieve peace. Especially now that peace has become so precious, and is in increasingly short supply.

Legal Layoffs

Lawyers and other legal professionals are trained peace-makers. This role extends into all facets of politics, business, and society. So the shift is not what we do, but how we do it. Currently, we’re seeing these tasks emerge in related areas:

Chief AI Officers (CAIOs) are gaining traction as artificial intelligence (AI) becomes more integral to businesses. A CAIO can help companies leverage AI to outperform rivals. Eventually, AI will build AI, so you don’t need to know what RAG is. What you do need is strategic regulatory and security expertise. Essentially, AI will replicate but will not regulate itself.

Customer Due Diligence (CDD) is a process used by financial institutions to prevent financial crime and uncover any risks. I overheard this during a radio broadcast: Dutch banks annually hire more fraud fighters than the government can hire police officers. Point: policing digital crimes is harder than we can handle.

Fact-checkers verify the accuracy of claims made in public discourse. Currently, lucky lawyers do the same while working with AI models trained for legal work. Their job is to double-check the outputs for inaccuracies. They are lucky because it’s now unaffordable for most. Eventually, legal professionals will be supervising inputs and verifying outputs from law bots. Remember in legal, hallucinations can also be a feature, not a bug.

Extra Time

Debatable studies showed that machines can negotiate and review contracts more efficiently than humans. Undoubtedly, machines draft faster. What about legal research and legal briefs? According to former president Obama, AI is already on a 4th-year law associate level. This is similar to what presidential candidate Al Gore said about eDiscovery over ten years ago. We have seen this movie before.

Inevitably, every traditional legal task will be done by a legal operating system. That’s what AI is: an operating system to perform actions at scale. The legal industry will develop its operating system, with special action models which can also run off-grid. Now imagine, who would we need to work on this?

To summarize: If sport is any indication, society needs about eight (8x) times more legal professionals than we currently have. We’ll also need a lot more automated legal arbiters to maintain peace. And they all need to be developed and sense-checked.

In closing: It’s disingenuous to keep saying to lawyers that they will always be smarter than AI, no matter how many trillion tokens it has. Legal professionals are the best-paid professionals on the planet. But only in London, New York, San Francisco, and other major metropolitan areas. And only involving very high-stake emotional transactions. So a lucky few get to play in the Champions League of Law. The rest of us will be running the streets, practicing the law we love.

Image Courtesy: DALL·E 3 by OpenAI using Glibatree Art Designer and the content of this analysis.

We welcome you to the weird world of GenLaw with our 1600+ legal GPT store. Now, ponder: Can we build Artificial Legal Intelligence (ALI)?

Subtext: Philosophically, we should build Artificial Legal Intelligence. Paradoxically, the question is: Will we ever be allowed to use it? Previously in Agents, we discussed the theoretical possibilities. Shifting focus, in this segment we’ll present the practical applications.

Three years ago, we made a video that foreshadowed the struggle. The video below shows how much money is invested to influence the law compared to how little to improve it. In 2020, we already pointed at OpenAI, among others.

This content is blocked. Please review your Privacy Preferences.

Reject / YouTube

Academic Angle

In the previous post, we stated that legal data has the complete DNA to create autonomous legal services. So by adding legal data and aligning Large Language Models (LLM), we should already have Artificial Legal Intelligence. However, we have yet to see ALI. Notably, here are some academic observations as to why we haven’t:

• Domain-specific large language models tend to underperform;
• LLMs perform worse on legal queries;
• LLMs perform worse when continually fine-tuning (catastrophic forgetting – CF);
• Excessive prompting causes ‘Butterfly Effect‘ and degrades the outputs.

I shared my views in the first and second study on LinkedIn and The Legal Tech StartUp Focus. Most importantly, my friend John Barker made another observation on GPT and Claude: they have been getting worse over time on specific legal queries. In this interview, Sam Altman acknowledges that out of the 10,000 answers an LLM will offer, maybe one of the answers will be perfect. GenAI relies on embedding vector databases, which results in a different output every time. Having 10,000 slight variations on the same topic will be a big deal in some areas like Health.

In Legal, we need consistency on every output. So using GenAI to dispense accurate legal answers every time will be challenging. Baffling to me: why can GenAI generate programming code consistently? Undoubtedly, progress will address these concerns.

Apps Aligned

Considering the above, have you tested any of the Copilots on Legalpioneer? I would love to hear your thoughts. Here are some, that may never get funded but matter to our everyday lives:

• Food Safety
• Pitbull Ownership
• Guns
• Visitation Rights

Contrary to the above, here are some topics that seem obvious candidates for funding:

• Security
• Human Resources (HR)
• Anti-Money Laundering (AML)
• Know Your Customer (KYC)

Of course, there is a large set of copilots looking at contracts in spaces like:

• Mergers
• Music
• Real Estate

We’ve seen legal copilots with multi-modal support and graphic outputs to help you visualize laws or patents. Some might ask: where are the apps for lawyers? We have research and drafting apps, but again, do we dare to use them?

APEX Apps

This brings us to the future. At the end of 2022, we told Law360 about the increase in consumer-facing legal. Now we’re seeing these client-centric apps proliferate. We will also see the emergence of apex apps. These apps aim to sit on top of the food chain. Perplexity AI and ARC Search are two examples.

Let’s recap: There is a blueprint for Artificial Legal Intelligence and a roadmap. We’ve posted the ‘road signs’ on LinkedIn. Here are five evolutionary stages we’ve seen, with the date we posted them:

1. GenAI is an operating system (OS) – January 11, 2023
2. The types of legal copilots – November 13, 2023
3. Likely most popular legal copilots – December 15, 2023
4. The power of Decentralized ActionAI – January 14, 2024
5. Likely most dominant copilots on law (Apex) – January 31, 2024

Point #3 had the second most views on my feed ever, and point #4 had one of the fewest. The above reminded me of this famous quote:

The future is already here – it’s just not evenly distributed

William Gibson

In closing: I made the video below on a beach in Portugal during my summer break in 2023. This blueprint was part of our September 23 webinar. It looked weird when I made it, it makes more sense to me now.

Do you have thoughts or concerns about building ALI or Copilots? Let me know on LinkedIn.

This content is blocked. Please review your Privacy Preferences.

Reject / YouTube

Image Courtesy: DALL·E 3 by OpenAI based on the content of this analysis.

We studied 4,172 mergers and acquisitions and distilled three insights on how founders exit their companies. So, what is peculiar?

This week we released our Mergers and Acquisitions (M&A) numbers. You can find the download links at the bottom of this post. In 2023, the most dramatic stat was the 85% drop in total transaction value. Furthermore, our analysis revealed M&A activity has been decreasing since 2023. Now, when you hear so-called experts say, ‘We’ll see more consolidation,’ remember: that is a lazy statement trying to look smart. However, despite the dramatic drop in transaction activity and value, one other peculiar trend caught our attention.

Acquisition Targets

First, let’s answer a straightforward question: Why would one acquire a legal tech enterprise? Looking at the 95 deals we registered, here are three insights that we could deduce from all:

1. Acquisitions are mainly driven by targets, talent, or technology, in that order;
2. Targets refer to customer bases that help boost the balance sheet with a robust ARR (Annual Recurring Revenue);
3. Talents achieve the best outcomes through bootstrapping, as any outside funding makes them vulnerable to lockups;

None of the points above should be shocking for anyone. Indeed, acquiring a company that has already done the hard work is the fastest way to gain a fresh set of attractive customers.

Consequently, companies accomplished this by either being a better seller or building a better product. This leads us to question whether this was a talent or technology play. Now come the handy insights for founders: did you really build the fastest car, or are you just an excellent driver? Because once a founder has been in the driver’s seat, it’s almost impossible to get comfortable as a passenger.

Acquiring Tech & Data

We continue our tradition of not naming names. Yet, we can manage two examples to illustrate what looks healthy to us.

3E, a provider of chemical compliance solutions, has acquired Chemycal, a 7-person regulatory monitoring shop based in Rotterdam. Additionally, Chemycal aggregates chemical and product compliance data from over 1,500 sources and maintains a library of substances and materials to keep customers updated. Thus, the acquisition combines 3E’s expertise in environmental, health, safety, and sustainability compliance with Chemycal’s global regulatory monitoring capabilities.

The private equity firm Thoma Bravo is acquiring the German software company EQS Group, which specializes in compliance and investor relations software, for €400 million. This acquisition price, marking a 53% premium over EQS’s previous stock price, indicates that the deal is likely to be completed. Despite EQS being a relatively small public software company, this move has led TechCrunch to ponder whether compliance tech represents a good startup bet.

Adding Debt

Ultimately, this brings us to the reason you probably clicked on this analysis: DocuSign. It was first reported on December 15, 2023, that DocuSign was working with advisers to explore a sale. According to the Wall Street Journal, it could potentially be one of the largest leveraged buyouts in recent times. While debating this on LinkedIn, one thing dawned on me: why private equity (PE)?

As always, there is an upside and a downside. Notably, Private equity firms offer founders and owners a generous way out of their businesses. The downside: a private equity acquisition is not the same as Facebook buying Instagram or Google buying YouTube. The Financial Times reported that private equity firms often resort to buying back companies after IPO flops. If you wonder why private equity buy companies, check our Power post. The TL;DR: private equity firms are investment vehicles that enable high returns for their investors. And they are able to achieve that by fancy financial engineering.

This perhaps explains why private equity firms acquire companies with little to no prior venture capital (VC) investment. After examining M&A transactions we have registered, it’s clear the majority of companies acquired by PE do not have any VC capital. Even when a PE-backed company goes on a shopping spree, they actively avoid VC-backed companies. Very peculiar.

You can download the M&A report here and schedule a chat for a one-on-one deep-dive.

We discovered our first legal agent on the first of November. First of December, we counted 559. What are Agents, and what can they do?

TLDR: Legal content is perfectly structured to assist Legal Agents. Legal Agents became possible by GPT fixing many of the downstream issues. Now there are three more obstacles on the road to Autonomy.    

Intro: This is our first-ever double header. The previous one was a bit dark, and we want to end the year on a bright note.

Ante

Before we throw any numbers at you, a bit of context. We covered a similar eruption of new apps twice before. In 2016, we recorded 44 new companies in a couple of months. We called it the Hottest Summer. During the crypto craze of 2018, we found 55 ‘legal tech’ companies raising a Billion in coin. Back in 2015, ProductHunt report, we uncovered a growth of 43 apps in roughly the same amount of days. In the early days of the Apple App Store, we noticed a similar growth. But nothing quite like this.

We were warned four months ago when we found 81 apps in 28 days. 550+ new solutions in one month set a new record. Now, will these new interfaces all be displacing current legal solutions and services? Not yet, and we’ll explain when this displacement will happen. These new apps will face similar struggles as the App Store, and ProductHunt platform posed. However, now there are different market conditions. In “Data, Distribution & Deals” we discussed market conditions that affect supply. This analysis will explore the demand side. Why do we need agents, and are they better?

Ancestors

In order to answer, here is a brief history of how we got here. We first discussed agents in 2011: Call My Agent. Basically, we wanted systems to learn from our content, derive instructions and tasks, and complete these tasks. They could perform their tasks autonomously in the background. This would allow us to access the results instantly at runtime.

In 2013, we created Loupe as a concept that would track regulations and instantly update affected contracts or briefs remotely. You would need a dashboard (DESH) to keep track of all edits deployed on each document. We even made an Apple Watch interface.

The same principle applies to legal research. We could dynamically convert phrases in documents to background queries. That was called Monocle and even made a teaser video. We submitted eleven patent proposals. Little did I know that each submission costs a fortune. Only one was patented, making me an inventor.

Why did we bring this up? Cooking these concepts in the lab, we discovered a few fundamentals of legal work. First, autonomous legal research is about anticipating legal issues, and calculating them beforehand. Second, the legal issues are already present in legal content. You just need to add the right context, as shown in the image above. In short: legal data already has the instructions and tasks baked in.

The biggest problem back then is actually what everyone now knows: calculating has immense costs. The OpenAI breakthrough is that no one has to pay that much. By creating GPT, OpenAI has democratized Legal Agents. Give GPT the right legal data, and anyone can create a Legal Agent.

Agents

What is a Legal Agent? Based on a given context, a legal agent can serve up a legal end-product like a contract, or research brief. If you ask GPT, it will provide a more technical answer. Something like “entities that can perceive their environment through sensors and act upon that environment using actuators. They’re designed to achieve specific goals or perform certain tasks.”. Not helpful, except it does provide the key ingredient for agents to work: context. Legal Agents get Legal context from instructions and legal data.

Before GenAI, every legal technology solution had to derive context from the user manually. As a Legal Knowledge Engineer, I had built legal decision trees and the process was very tedious. The tools were cumbersome and we could never encompass all cases. We just focused on the most complicated ones for commercial purposes. Now, with the right instructions, and a fined-tuned model can figure it out. Better yet, you can ask to explain the answer. This was the real breakthrough of TaxGPT. So are we there yet? Not yet. Let’s highlight one key difference we uncovered while covering GPTs: Copilots answer your questions, Agents take actions. So what is the next stage?

Autonomy

The launch of GTPs on November 6, 2023, made instructions more explicit for ChatGPT. However, OpenAI demoed instructions months earlier on March 14th, 2023. Remember TaxGPT? Yes, the first agent by OpenAI was targeted at Tax. Now there are around 90 TaxGPTs and counting. However, none can submit a tax form to your local IRS. While the current generations of legal agents can give you perfect answers, most can not take any action. Part of this constraint is security by design. Large Language Models will prevent any kind of code execution outside a sandbox. However, developers, that have APIs for actions ready in their apps, can now use GPT output as input. As the Copilot Battle states, this is the footrace, so hurry. Wait! There is one more thing.

Besides structured legal issues, we made one more discovery in our Ancestor era. A quick step back: The best definition of Artificial Intelligence I heard here at 4:59 min on the Waveform Podcast. The short version: AI is planning tasks in the right order to execute them. Sticking with the TaxGPT, here is a simple plan:

1. Upload your bookkeeping*;
2. Classify entries;
3. Find the right deductions;
4. Fill out the tax form and;
5. Submit tax form to Tax Office.

*Note: Another Agent did my bookkeeping.

What tax and legal data provide is an understanding of how to structure tasks and actions to solve legal problems. Legal data has a complete recipe. Given this recipe and the ability to take action, Agents will be able to act autonomously. So the next stage of legal solutions is autonomy. And now you know why Sam was fired.

Agree

On the roadmap to autonomy, we have to travel through the Three Universes from “Will lawyers be replaced by GPT” which are:

1. All legal data needs to be processed in GPTs. As stated in the previous post, it eventually will;
2. Learn how to construct Agents to allow them to safely take actions, so allow autonomy;
3. Interceptors do not squeeze developers, so allow fair competition;

This third milestone is, economically, the most difficult one because of Interceptors. Example: Google Bard now supports extensions and they are all Google apps. OpenAI was supposed to be open, but now it’s just Microsoft Copilot. Ideally, GPT becomes an open utility, just like GPS. Politically, it seems we are beyond that possibility. Potentially, agents can save the world, if we agree we want them to.

Meanwhile, we counted over 24,000 GPTs created since November 6th. That is 750 GTPs a day.