Every billion-dollar company enjoyed a period of growth, growth protected by a moat. Here’s the blueprint for a moat in legal.
Fellow legal geeks, here’s Legalcomplex’s last transmission. I’ve brought you legal trends from different angles, always aiming for an original approach. Spotdraft’s CEO Shashank Bijapur called this: taking a telescope and microscope to the legal market. His and others’ inspiration fueled each of the 139 posts. Let’s go out with a bang: here’s your billion-dollar blueprint.
Profits.
Law firms will likely never again find a business model as lucrative as hourly billing for legal advice. AI will replace traditional legal work in many spaces. Meanwhile, AI will become the largest cost center for nearly every business as it becomes their new operating system. The system that codes new solutions, or summarizes long legal documents. The actions and answers underlying our decisions. Since any high-value decision has to consider the legal ramifications, legal will forever live on.
So is there a billion-dollar business hidden within the legal market? Yes, one that leverages moats to protect valuable decisions. Business moats offer a company a sustainable competitive advantage, protecting its profits and market share from competitors and external threats.
What are the business moats in the legal market? In Data, Distributions & Deals, I explored the moats of the biggest legal tech companies. For law firms, moats are pretty straightforward: proximity to Capital and Conflicts. Since Roman times, legal counsels were the highest-priced service in society, with pricing set by law. So lawyers have historically protected legal profits by law. Are these profits still protected?
Protected.
What’s up with Devin.ai? Remember the math prodigy who raised $196 million at a $2 billion valuation? Well, Cursor.com happened. The slideshow below shows the difference in their respective approaches. Now neither Devin nor Cursor have a strong moat. OpenAI’s Codex copied the best features of both, except one very important one. Put a pin on this. Contrary to common sense, having killer features only allows you to be killed. No matter how many models we orchestrate around them. Especially cloud ones.
Every technology disruption generates a bubble where the first wave fails. As illustrated in my Bubble post, investors’ initial picks usually are not the winners. Additionally, I missed mentioning a third bubble: 2017-18 ICO Craze. The Initial Coin Offering (ICO) collapsed all applications, except for one: stable-coins. The two largest stable-coins, Tether (USDT) and USDC by Circle.com, collectively now account for $200 billion in value. This value largely stems from the use of stable-coins in payments and remittances. However, it took Blockchain seven years to find a viable business model that yielded durable, profitable revenue.
‘Durable’ implies that their growth is protected from competitors by a moat. Having a substantial war chest of capital to fuel growth also does not offer any guarantees. Microsoft is laying off over 6,000 employees. Here’s another mystery: 1,120 employees disappeared on the OpenAI LinkedIn profile page in 79 days. I stumbled on this little detail while processing their latest fundraise. Check the final two slides below. These signals should not surprise anyone, we’re simply watching the replays of the past. A past where cloud was king.
Privacy.
In ancient times, the powerful always enlisted the counsel of confidants to make key decisions. Usually, this happened in secret to avoid the influence of peer pressure. Today, core constitutional rules for any democracy include the right to cast a vote in secret and the right to independent legal representation and an impartial judge or jury for a fair trial. The key to making correct decisions is having access to unbiased information. To make any information unbiased, one must balance these three sources of data:
- Public
- Private
- Personal
My reason for visiting Alexandria, Egypt, was to design secure Legal AI for judges. To ensure their legal decisions could not be influenced. Sabaio understood this principle at a core level. Check the row ‘Integrations’ in the Devin vs Cursor slide below, you’ll notice that Cursor is ‘local-first’. That is the moat OpenAI can not copy. Unpin.
This brings me to the only moat that legal actually has: Trust. Trust is earned by keeping communication confidential. Storing information on-site and not sharing it with third parties. That AI inputs and outputs can not be part of breaches at providers or litigation by competitors. In case you missed it, OpenAI was ordered by a judge to retain all outputs. As insane as this may sound, we should remember that AI is running out of data. Thus, synthetic data based on human inputs is invaluable for reinforced learning (RLHF). Remember my viral post on RLHF and sacrificing your queen?
Price.
Let’s be realistic. In End of an Era, I explored the economics behind bringing privacy to enterprises– my sacrifice. In a recent IBM survey, the costs for companies to host AI and hire talent were identified as a hindrance for many. Nevertheless, what some call “sovereign AI” does seem to be a goal worth investing in. My two notable examples are Mongolia and the Middle East. Mongolian startup Egune AI raised $3.5 million to ensure “..native languages and identities are preserved and empowered in the digital age..”.
Life would be pretty boring if we ate hamburgers every day.
Let’s wrap this up. Legal has one moat, and that moat is trust. Trust is earned by upholding confidentiality. In the age of the internet, confidentiality does not come cheap. If you crack the code to offer affordable confidential computing, you will ride a wave towards a billion-dollar revenue run rate. The key is to build a business model that packages hardware and software at an affordable price, a price competitive with most cloud offerings. That is the blueprint, get cracking. End transmission.
