Robin AI Launches GenAI Due Diligence Reports in Major Move

The world of legal genAI has had an unmet need – how do you handle M&A due diligence, which can involve tons of lengthy documents? Now, Robin AI has met the challenge and launched Robin AI Reports to help with this.

This matters a lot, as it was M&A due diligence where several of the first legal AI companies in the previous decade got started, as due diligence is laborious and slow, and AI looked perfect as a tool to help. However, while many lawyers are today using genAI, leveraging the tech for major transactional review work has not been so well developed. So, this could change things.

Through the Robin AI platform, lawyers can click to generate a report which is emailed to them on completion. Users can then build a list of ‘red-flag’ issues, and deviations from preferred positions, which they want summarized for each contract.

Robin AI Reports then provides ‘an accurate summary of the issues for each contract, with clear citations, so that the summary can be easily verified by a human. The easy-to-read reports can be generated for a wide range of purposes including M&A due diligence, NDA compliance, supplier agreements and audits’, they explained.

Plus, the company is allowing users to access three reports for free to get a sense of how it works.

They added that Robin AI Reports was developed with partners including the University of Cambridge’s Investment Management group, who have been using the product since April.

The group found that ‘a task that used to take three hours to complete by Cambridge Investment Management’s senior in-house lawyers now takes 30 minutes on average – a roughly 85% time saving’. And that’s impressive.

Moreover, its use of Amazon Bedrock allows Robin AI to leverage Anthropic’s Claude 3 generative AI model alongside Robin’s own models, while ‘ensuring customer data remains secure and compliant with local data regulations’. A copy of Claude 3 exists inside Robin AI’s cloud environment, meaning user data never leaves Robin’s cloud.

Part of an example report.

All well and good. AL asked Robin AI CEO, Richard Robinson, some additional questions:

If this is for M&A Due Diligence how many documents can you ingest at one time? 

Several hundred for now. We recommend only loading individual documents 300 pages or fewer in length, but we’re working hard to expand this capacity rapidly.

What is the cost like?

We’ve structured this on a usage-based pricing model, so that customers only pay when they use the product and get value from it.

From your testing so far, how is the accuracy?

There’s nine different document types and we’ve been hitting 90% accuracy from internal testing and benchmarking – for example for citations offered in reports on supply agreements and shareholder agreements.

We believe that lawyers still need to be in the loop when using a product like this. We just urge users to remember that neither AI nor human lawyers can operate error-free.

This is where the citations are crucial. They speed up lawyers to quickly go back and forth between the summary and the actual contract, to verify information.

How big is this in terms of growing Robin AI?

When you consider that due diligence is probably the biggest revenue driver for law firms across the world, this product aims to massively accelerate it. I think it’s our biggest product launch to date. In five years’ time, every lawyer in the world will be using an AI product like this.

James Clough, CTO and co-founder of Robin AI, added: ‘With Robin AI Reports, M&A, IPO, and other asset transactions no longer have to grind to a halt for weeks for due diligence. We’re helping businesses close deals faster and are proud to be the first in our industry to do this.’

Robin AI now has 140 full-time employees and over 100 customers. They raised $26m in Series B funding in January.

Is this a big deal? Yes. Ever since genAI arrived people have been wondering when due diligence for M&A – and as they mention for other use cases – would see a reliable tool using LLM technology. As noted, due diligence lends itself to automation because of the high volume of documents and the fact that both lawyers and clients see it as a painful, but essential step for any deal. If this can be made faster and more cost effective for the clients, then there is clearly a market for it.

Older NLP machine learning tools are still used for this work, and do well at it, as they’ve been trained for years on this type of matter. However, genAI could be used here as well – but scaling challenges for large and multiple documents have acted as a barrier.

Plus, it’s worth saying that the legal tech grapevine has been whispering about Harvey launching a due diligence capability of its own as well. Looking forward to seeing how this develops. It’s certainly going to be a new field of expansion for legal genAI.

Congrats to Robin AI on the pioneering work.