The Failure Of Atrium, A Promising Law Tech Firm Proves That AI Software Is Not A Panacea

“Software is eating the world”, proclaimed Andressen Horowitz, a leading venture capital firm in Silicon Valley which invested in Facebook, Box, Lyft and many other leading technology companies. The story of software, especially AI software invading the world and replacing physical processes with digital ones. With Uber and Ola, software ate into the world of transportation, with the advent of Netflix and Youtube software ate into the world of entertainment, and we can go on and on with such stories. Such stories are omnipresent. 

But many have also questioned how long this story and the trend continue. There are definitely some snags in the trends. The overall thesis of taking an industry and introducing AI and software, and seeing massive rises in productivity and output is suspect to many real-life constraints. One of the stories that have come to light in the last few days is Atrium – a promising law tech startup bringing AI and software automation to the legal world. Atrium recently announced its closure while returning some of the money they had raised from VC and laying off around 100 full-time employees.

The startup was founded by Justin Kan, a second-time founder and a former partner at YC. He had previously co-founded Twitch which was acquired by Amazon for nearly $1 billion. Kan took on the challenge of building a startup that will deliver more efficiency than a traditional law firm using – you guessed it right – software and AI. The stage was set for massive success with all the correct ingredients, maverick repeat founder, an old industry with legacy practices, the tech-focused vision of the founding team, the backing of many of the best VC firms in the world. What could go wrong?. Now many experts are puzzled and inspecting exactly what went wrong.

Domain Expertise In The World Of AI

Let us make one thing clear. There are a few founders who are better than Kan. He has been an exceptionally successful technologist with an impressive background. The question of where Atrium failed has to not only answered by Atrium but also the legal tech community because therein lie the necessary lessons for the future. Justin Kan even though an impressive founder was not a domain expert or even someone who came from the world of law. 

Companies offering a full suite of AI-augmented services have grown considerably in the recent past. But in some domains such as law, medicine and others where regulation and licensing plays a huge role, some very important issues arise. In many countries, court cases and issues regarding AI agents practising law or legal activities have come up. These can be called UPL issues (unlicensed practice of law), where software is performing actions some qualified lawyer should have been performing. 

Same issues might crop up in medicine, so many healthcare and medical startups make sure to advertise their solutions as an augmentation to the medical professional rather than replacing the practice. There were also some complaints from Atrium customers that pointed to post-sale service, that did not match the service provided by the legal industry. 

Startups entering the legacy service industry in the fields of law, medicine and others might do well to remember the value of service lies in the quality of the service. In the current age, AI has not yet reached a level where it can match human service experience at scale. We will reach that stage but it is definite that we have not reached there yet.

The Bet That Did Not Work

Atrium started with inhouse lawyers with intelligent software to assist them initially. In time, the lawyers were to be automated and margins would improve. So back in January, Atrium began to layoff lawyers on the staff and convert them to freelancers. They would create a network of legal service providers and would now to pay them a fraction of the amount compared to paying legal experts full time. This bet didn’t pan out exactly as planned. The service experience for customers was not as good. Many of the clients left with the legal expert rather than rely on a firm providing more software and less legal expertise. To make matters worse many clients of Atrium didn’t exactly know who their legal representatives were.

The bet that many startups are willing to make nowadays is that they can run full-stack operations like a traditional company while adding software and AI to make it more efficient. Turns out full-stack companies are better served by startups when they are offered software playing only on some level of stack helping them maximize their efficiencies internally.

A good lesson for companies here in India to keep in mind too. AI software is not a panacea for any legacy sector that can make it all efficient and create bountiful opportunities for the entrepreneur. 

Download our Mobile App

Abhijeet Katte
As a thorough data geek, most of Abhijeet's day is spent in building and writing about intelligent systems. He also has deep interests in philosophy, economics and literature.

Subscribe to our newsletter

Join our editors every weekday evening as they steer you through the most significant news of the day.
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.

Our Upcoming Events

15th June | Online

Building LLM powered applications using LangChain

17th June | Online

Mastering LangChain: A Hands-on Workshop for Building Generative AI Applications

Jun 23, 2023 | Bangalore

MachineCon 2023 India

26th June | Online

Accelerating inference for every workload with TensorRT

MachineCon 2023 USA

Jul 21, 2023 | New York

Cypher 2023

Oct 11-13, 2023 | Bangalore

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox

Is Sam Altman a Hypocrite? 

While on the one hand, Altman is advocating for the international community to build strong AI regulations, he is also worried when someone finally decides to regulate it