Patent Work Automation – Just a Dream?

What comes to mind when you hear the word “automation”? The Industrial revolution 150 years ago? Robots at factories? The IT boom of the 80s and 90s? Same here! 

In recent years, though, I’ve been focused on a different kind of automation – teaching AI to think like a patent professional. I was reminded recently that my first blog post for IPRally, back in 2018, was titled "Patent Automation – It’s About Time”. Beyond the clever wordplay, it reflected a genuine belief that we could significantly streamline the work of patent professionals and, in some cases, even remove humans from the loop entirely.

After 6 years, it’s time for some self-reflection. 

Have we achieved the goal yet? No! But we’ve been building a strong foundation for it.

Have we lost the belief? No! With all the technology developed by us over the years and available from others, this isn’t the time to lose sight of the goal. 

Who would have believed even two years ago that AI would be analysing, fixing, writing and running program code for relatively complex software? Or creating high quality images and videos by just asking? 

Sometimes things take time. Sometimes it’s hard to follow the pace. Ironically, for the automotive industry, it’s taken over 100 years to get to the auto-motive, i.e. self-driving, state, and it’s still a work in progress. The challenge there is comparable to the patent industry: freeing up human drivers from important, but still somewhat routine tasks to more productive work, while reducing mistakes.

IPRally’s vision

We have re-set our goal to automate repetitive substance-level patent work in a transparent, verifiable and trustworthy manner.

There’s lots of sophisticated words there. So, let’s take a closer look at what we mean (and don’t mean) by them. 

• Automating work means that we’ll utilise AI even more and push its boundaries even further, with an end-to-end approach. That is, the focus will be on solving challenges across the patent process from the initial IP question (e.g. “Is X patentable?” or “Can we enter market Y with X?”) to provide the necessary information for making a reasoned decision. The aim is to reduce the turnaround time of such tasks from days and weeks to hours and, in some cases, even minutes.
• Repetitive work means that we focus on tasks that have a well-established purpose, flow and end goal within the patent space. This includes the different variations of prior art searches, freedom to operate and evidence of use type searches as well as some portfolio analysis tasks. This doesn't mean replacing all human work but rather automating the tasks that are best suited to AI. In doing so, we aim to elevate the work of patent professionals by improving productivity, quality, and consistency.
• Substance-level work means that we’ll focus on the technical content of patents, not processes such as IP management or docketing. This is perfectly in line with our patent specialist graph AI, which has been trained from the ground up specifically for this purpose. We are confident to say that it’s the most accurate patent search model in the world and the foundation for end-to-end automation efforts.
• Transparent, verifiable and trustworthy means … exactly that: Making the automated flow, including its limitations, understandable. Making the findings and reasoning of the AI visible and simple to review. Keeping the confidence level high enough with regard to the purpose of the task.
• Patent work means work that utilises patent information in one way or another –  whether it be the end user, a patent professional, an R&D engineer, or a C-level executive. According to our mission, we’ve always wanted to bring the relevant information effortlessly to the right spot in the value chain – this is usually as early as possible. 

We don’t intend to achieve all of this overnight. Instead, we’ll begin with common use cases and industry needs, gradually expanding from there. Just like our current core product, this will involve many baby steps and a lot of trial and error. We’ll rely heavily on feedback from industry pioneers along the way.

Gradually, IPRally will add AI assistants that solve parts of the process, and eventually, we’ll connect them to perform complete tasks in the value chain. 

And just like putting a man on the Moon – or even Mars – there will be many valuable discoveries along the way.

Some final dreaming

The power of AI in IP is to enable high-quality decisions to be made at the right time and at costs that are in proportion to its purpose. The AI cannot take responsibility for the decisions, but it can help to find, refine and help to understand the information for the human decision-makers. 

AI can even advise on what the decision should be, but as long as the AI doesn’t pay the fees or sit in the courtroom, it cannot replace the professional. What it can do, though, is help the professional become a better professional, or turn a good IP process into an exceptional one. 

We are seeing signs of this already with early adopters of AI tools across multiple use cases. But honestly, the big transformation is still to be seen.

The automation challenge is a big one. As a patent attorney, I know that by heart. No single technology is ready out of the box. A lot of development is needed. We’ll see some overpromises, too. But I don’t blame people for their optimism.

I also know that search automation is exactly what I wanted when I sat by my desktop work PC and landline phone (remember those?), getting frustrated when receiving a foreign office action with a killer piece of prior art, 3 years after filing a patent application. The process felt somewhat outdated, even 15 years ago.

Maybe one day soon, I’ll be able to say: “Remember when you had to spend days on a patent search?”

In my upcoming blogs, I will explore the technological, human, and organizational perspectives while shedding light on our roadmap.

If you read this far, I’d love to hear your thoughts on the topic. Feel free to reach out to me through Linkedin, or by emailing me at product@iprally.com.

‍