Hello and welcome to Eye on AI. In this newsletter… why a legal AI startup shows the AI ​​boom is about more than just basic models; Zoox begins offering robo-taxi rides in San Francisco; it is extremely easy to jailbreak LLM-powered bots; is the progress of the foundation model at its maximum?

There was a lot of talk last week about whether advances in general-purpose foundation designs could be hit a wall and what it means for the AI ​​boom. (More on this in the Brain Food section below.)

Some skeptics, like Gary Marcuspredict a toll comparable to that of the Internet crash. I’ll be discussing this topic tomorrow at Web Summit in Lisbon, moderating a center stage conversation at 4:25 p.m. local time on “Is the AI ​​bubble about to burst?” with Bhavin Shah, CEO of Moveworks, and Sarah Myers West, Co-Executive Director of the AI ​​Now Institute. You can view it on Web Summit Live Stream!

My view is that even if progress on the base model slows, it may not matter as much for companies implementing AI applications for specific industries as it does for companies such as OpenAI whose $157 billion valuation appears largely based on the achievement of artificial general intelligence (AGI). Or at least it relies on a scenario in which OpenAI remains at the forefront of model development and has some sort of defensible moat around its business, which won’t be the case if building LLMs still larger ones does not confer a significant advantage in terms of capabilities. justify the cost.

These are solutions, not models

Many of these AI application companies are in the business of selling a solution to a specific industry problem, not selling a particular AI model or a vague concept like “general-purpose intelligence.” In many cases, these solutions do not require AGI, or even necessarily additional advancements in AI capabilities. In some cases, simply coupling several existing models and refining them on data relevant to a particular business task is enough to create a pretty good business.

A good example of this in the legal tech world is Robin AI. The company was founded in 2019 by Richard Robinson, a former lawyer at Clifford Chance, and James Clough, a former machine learning researcher at Imperial College and Kings College London. Robin doesn’t just sell companies a particular technology. Instead, it sells legal services to large corporations, with some of those services provided automatically through AI software, and some of those services provided by Robin’s employed human lawyers and paralegals, who are assisted by technology, including AI, that Robin developed.

“It’s about doing things that models are currently capable of, but also investing in what’s simply out of reach today, and then using humans in the loop to fill the capability gap.” , Robinson told me.

Closing the expectation gap

He recognizes that “there is a gap between what people expect from models and what they can actually do reliably.” For example, he says, the most advanced AI models are now great at synthesis and pretty good at translation. But they cannot yet reliably negotiate a complex legal document or accurately draft a brief for a court case. “They can do part of the task, but nothing like the whole thing,” he says.

But – and this is the bottom line – Robin AI has a viable business even with these shortcomings. And its business will remain viable even if these gaps close only slowly, or never close at all.

Indeed, while some clients simply purchase the software from Robin, others outsource an entire legal task to the company – and it’s up to Robin to find the best way to accomplish that task. at a given price.

“We have staff, but they are highly optimized with our technology, which significantly reduces costs,” says Robinson, emphasizing that the company does not engage in labor arbitrage by hiring paralegals in low-cost countries like India or the Philippines. Instead, it employs lawyers and paralegals in New York, London and Singapore, but they can work much faster thanks to Robin’s legal co-pilot technology. And this technology doesn’t just consist of core models developed by OpenAI, Anthropic, and Meta, but also a host of other technologies, including search algorithms and old-fashioned hard-coded rules, all strung together in a complex. workflow.

A “B Plus” round of $25 million for Robin

In a sign of confidence in Robin’s prospects, Eye on AI can announce that the company has closed a $25 million “Series B Plus” fundraising round, in addition to its initial $26 million fundraising round. series B. announced in January. This brings the total amount raised by Robin AI to date to $61.5 million.

Investors in the new funding round include PayPal’s venture arm, billionaire Michael Bloomberg’s family office called Willets, and the University of Cambridge, all of which are also Robin AI clients. The original Series B round was led by Temasek. The company did not disclose its valuation following the latest investment. He said he currently earns $10 million in annual recurring revenue.

Robinson says the company wanted to invest more, even though it still has a lot of financial headroom from the initial Series B, in part to add additional features to a product called “Reports” that has proven particularly popular with customers. Reports allow users to ask an unlimited number of questions about a set of documents. He uses Anthropic’s Claude model under the hood to power his answers. Robinson says the company hopes to add even more reasoning capabilities to what Reports can do, but using the most advanced basic models increases the company’s costs, which is why it’s helpful to have additional funds in the bank.

Robin AI also competes with many deep-pocketed rivals, including Harvey AI, which is backed by OpenAI and which raised a $100 million funding round last summer at a $1.5 billion valuation . It also competes with products from Thomson Reuters, which owns Westlaw and has acquired several legal AI startups, including Casetext, which it bought for $650 million in 2023.

Analyze 10,000 contracts in hours

In one recent case, Robin claims to have helped an unnamed US biotech company deal with a data breach, by reviewing 10,000 contracts, spread across 30 different contract types, to understand what the biotech’s obligations were in terms of information of the counterparties of the violation. Using the Reports product, along with Robin’s human legal experts, Robin says the biotech was able to identify the 50 highest priority contracts that required notification in just a few hours, and have an action plan in place for all 10,000 contracts within 72 hours. He estimated that this would have saved the biotech company 93% of the time and 80% of the estimated $2.6 million it would have taken to hire an outside law firm to manually review contracts. This is the value that businesses are getting from AI today. And its value won’t disappear, even if GPT-5 isn’t as big an advance over GPT-4 as GPT-4 was over GPT-3.

With that, here’s more AI news.

Jeremy Kahn
[email protected]
@jeremyakahn

**Before you get the news: If you want to learn more about what’s next for AI and how your business can get ROI from the technology, join me in San Francisco on December 9-10 For Fortune Brainstorm AI. We’ll hear about the future of Amazon Alexa from Rohit Prasadsenior vice president and chief scientist of the company’s artificial general intelligence; We’ll learn more about the future of AI generative search at Google through Liz ReidGoogle vice president, search; and on the form that AI will take Christophe Youngexecutive vice president of business development, strategy and enterprise at Microsoft; and we’ll hear from the former San Francisco 49er Colin Kaepernick about his company Lumi and the impact of AI on the creator economy. You can view the agenda and apply to attend here. (And remember, if you write the code KAHN20 in the “Additional Comments” section of the registration page, you will receive a 20% discount on the ticket price, a great reward for being a loyal reader of “Eye on AI”!)

This story was originally featured on Fortune.com