Juro’s vision for intelligent contract automation

We started Juro to help the world agree faster. 

Frustrated by the absurdly slow process of taking a contract from inception through to agreement, Pavel and I thought there must be a better way. We set out in 2016 to build a flexible, collaborative platform that anyone could use to process a legal contract end-to-end in their browser.

Fast forward 8 years and we have helped our customers process 2 million contracts in over 85 countries. Some of the world’s fastest-growing businesses, like Deliveroo, Remote, and Trustpilot have relied on Juro to help them streamline routine contracts at scale across multiple years of their growth.

We’re proud to have helped them, and many more, collectively save millions of hours on low-value work, accelerate deal closing times by more than 75 per cent and keep track of contract renewals amounting to billions of dollars.

But our story is just beginning.‍

What generative AI will do to CLM

Generative AI will bring the single biggest change to how we work since the Industrial Revolution. All teams within the enterprise are beneficiaries of this shift. However, legal work especially lends itself to assistance from AI. Indeed, according to Goldman Sachs, 44% of legal tasks will be automated by AI in the coming years.

Whatever the exact figure is, they’re basically right. When I was a trainee lawyer, in a windowless room searching for change of control clauses in a pile of contracts just a few years ago, I could not have imagined technology as powerful and as useful as the Large Language Models (LLMs) available today. I needed it then. I need it now.

We believe generative AI will fundamentally change how businesses and people agree contracts for three reasons:

1. Unstructured vs structured data

Contracts by their nature are unstructured. The valuable business information they contain is wrapped in paragraphs of legal text. “Whereas the aforementioned…”, “subject to the provisions hereof," and so on - you know the story. In Juro we have addressed this challenge in the past through ‘smart fields’: structured data points that extract key information from contracts so it can be reported on.

LLMs like GPT4 are great at extracting key information without manual work, which makes the distinction between data types much less important. AI will get you whatever you need from the documents instantly and without effort.

2. Simple task automation

Generally, in-house legal teams are small relative to the size of the businesses they support. According to our 2024 State of In-house Survey, 67% of in-house lawyers say they are ‘drowning in low-value work.’ This includes tasks like contract review exercises, summarizing documents and marking up endless NDAs. This was consistent in our surveys in 2020, 2021, 2022, 2023 and now 2024. Co-pilots built on LLMs can help you get through the grunt work faster. This can only be a good thing.

3. Task sequences

In-house lawyers spend much of their time creating and amending text. Think about drafting a term sheet and then taking the key provisions and turning them into long-form legal documents. Completing these chains of tasks is a highly skilled job, but the process of getting there usually includes a lot of low-value work on the way. Agents and agentic systems based on LLMs can do these sequential jobs for lawyers at ~0 marginal cost and to the standard of a Junior Associate. They can fully automate tasks, leaving the lawyer to review and amend the output.

Good lawyers have nothing to fear from this shift. No one went to law school to do low-value tasks. Rather, legal jobs will focus on the high-value work lawyers trained for; as a result, legal jobs will become more impactful and more fulfilling. Those who adapt will thrive; those who do not will struggle.

The first generation of CLM

One of the most curious parts of our journey is that when we identified the root cause of the problem with contract workflow we found a huge number of legacy solutions in market. Buyers understood that pushing contracts from Word to email to Slack to email to PDF to DocuSign to Sharepoint slows businesses down. They got it. They were turning to CLMs to find a better way. CLMs were everywhere then. They are everywhere now. 

But in spite of billions of dollars invested by VCs in CLMs, the largest of them are scraping $200m in recurring revenues and most are in the low single-digit millions. Compare that to CRM, where Salesforce on its own turns over $9.1bn. There is no Salesforce for legal and CLM as a category has made only a tiny dent in how contracts are agreed and managed.

Why is this? The first generation of CLMs suffer from being:

1. Centralized

Often built by lawyers for lawyers, many CLM vendors think that folks in the business want to engage with CLMs the way lawyers do. They don’t. 

Building CLMs around a central interface may work for lawyers as they work through their to-do list, but it creates a huge barrier for adoption by end users. If you need a contract, you really want to get it without leaving the tools you live in every day. CLMs have failed in both the breadth and depth of their integrations they offer, meaning folks in business teams generally don’t use them.

2. Uncollaborative

It’s 2024 but almost every CLM has been built around MS Word, a technology founded in 1983. Yes, you can argue that CLMs should ‘meet lawyers where they live'. Problem is, most of the users of contracts aren’t lawyers. They’re folks in Sales, HR, Finance and Procurement who rarely, if ever, use Word. And their expectation is that dealing with contracts should be as easy as collaborating in Slack, Teams, Monday or Notion. Lawyers might want to keep Word; but modern business teams want multiplayer collaboration in the browser.

3. Data-poor

Legacy CLMs require lots of manual work to get hold of data. Those of them that have deployed AI for data extraction often relied on first-generation AI. The problem with this is two-fold. First, lawyers rightly expect very good accuracy when it comes to capturing key contract obligations. That generation's AI was not accurate at all. Second, the range of data points that first-generation AI could capture was dependent on the training sets given to it by the user. This required a ton of manual work and the range of legal data points you could pick from was often not enough for your specific business needs.

50% of CLM deployments fail

All of this has caused, by WCC estimates, some 50% of CLM deployments to fail. They fail not, as many vendors state, because the customer doesn’t know what they need. They fail because the tools available don’t solve the root cause of their challenges in a way that folks want to use. This is a shocking statistic when you consider the investment that has gone into this category. It requires a fundamental rethink of what a CLM should be.

The next generation of CLM‍

We believe that generative AI can unlock much better solutions in CLM. We’re calling the next generation of CLM “Intelligent Contract Automation”.

Aided by AI, lawyers will end up executing fewer tasks in CLM. They will shift from being the front-end users of CLM to being back-end users. What I mean by this is that if lawyers can set the right playbooks and guardrails for contracts, there is no reason why folks outside of legal cannot self-serve on contract tasks.

Imagine a month end where lawyers are brought in only on the most critical legal points, with almost all deal advice and negotiation on AI autopilot. Lawyers like me don’t get much pleasure from addressing the same points again and again in an impossibly short timeframe. With AI, properly and safely deployed, we won’t have to.

CLMs will become intelligent. CLMs are platforms and that means that they are stuffed full of data. In Juro we have somewhere in the region of 10 million smart fields alone (and that doesn’t include any of the unstructured data in contracts). AI-native CLMs will have a contextual awareness of the customer’s contract stack. That means that a customer’s own data can be used to inform the output of AI. Your CLM will learn what you want, and do it for you.

Finally, contracts will be much more deeply embedded in workflows. There are integrations and there are integrations. CLMs need the latter. Imagine contracts that live entirely within Salesforce. Where a sales rep can get a contract in one click, negotiate it with AI, and get it signed - all within the tool they live in every day. Imagine needing a new employment contract, calling up a legal agent in Slack, and having a draft served back to you instantly. CLMs will become APIs - an interface to legal teams.

Towards intelligent contract automation‍

Juro has already taken strides to incorporate generative AI into our workflows. AI Assistant is helping customers draft, summarize and review contracts and extract complex information from contracts after they are signed. But more than just building AI features, we are structurally reimagining CLM for the AI era.

Our vision is to be the first truly intelligent contract automation platform. The key to this is not only building an AI-native platform, but in building tools that accelerate collaboration between lawyers and the business teams they support.

Our vision is built on three pillars:‍

1. Self-serve from where you live‍

Juro will empower all teams to process the full lifecycle of contracts from the systems they live in, reducing the time everyone spends on contract admin. Juro should serve as an interface between legal teams and the internal stakeholders they serve. This will boost trust in the legal team, reduce low value work and improve relationships with the business.

2. AI-native workflow

‍We will continue to power the end-to-end contracting process but identify and execute on AI-native solutions to jobs throughout the contract lifecycle. Legal will set the parameters and guardrails of what can be done and what cannot, and business teams will be enabled to execute contract tasks. This will dramatically shorten deal times for commercial contracts in particular but will help all types of contracts get over the line faster.

3. Intelligent repository

‍Juro will continue to be a system of record for all contracts, but will also enable users to surface contract data instantly - whether unstructured or structured - and act on it. Tasks like due diligence and contract reviews, which used to take weeks can be conducted in seconds from the tool in which your contracts live. And the more contracts you add to the repository, the more value you can extract from the legal data it contains. 

So what next?

The shift to AI in legal is happening already. According to our 2024 state of in-house report 86% of in-house lawyers have used generative AI and 44% are using it weekly or daily. This is a dramatic acceleration from a year ago.

Many point solutions are coming to market, often as add-ons to MS Word. In our view, CLMs are better placed to serve as the canvas for lawyers to use to automate contract tasks but they must adapt and fast. We have a unique opportunity to transform CLMs from clunky, unloved, poorly-adopted tools to centrally important platforms that make the lives of lawyers and the teams they enable better and deliver real value for businesses.

We are already on our way to realizing this vision. Things we did not believe possible when we started are now an indispensible part of our customers’ workflows. Things folks said would never happen, have happened. 

By delivering on the promise of intelligent contract automation, giving legal and business teams the tools they truly need to agree and manage contracts seamlessly we will speed up contracting times for them. Deals negotiated in minutes, not weeks; partnerships agreed in real time; new employees secured the same day. This in turn will help us achieve the mission we set out on 8 years ago: to help the world agree faster.

CLM is dead, at least as we know it. Intelligent contract automation is the future. 

Help us build it.