By Dannon Allbee, vice chair of patent group, IP practice at Potomac Law Group
Law firms are facing operational pressures, particularly in patent work, such as constantly increasing volume and rising complexity. Available time and budgets haven’t kept pace with these demands. Potomac Law Group’s (PLG’s) objective is to maintain proactivity while leading AI initiatives, rather than chasing the market and being forced to adapt. Legal AI tools have helped PLG increase efficiency without sacrificing quality. By leveraging AI to accomplish specific tasks, our attorneys can focus on strategic thinking without having to craft every sentence individually.
Most AI legal tools are built on a similar set of back-end large language models (LLMs). However, the results of these tools often felt generic, not IP specific, and certainly not fine-tuned to draft patent applications or office action responses. Therefore, we had to be very careful in drafting prompts for AI tools and figuring out where to fit them into our workflows. We often had to change our workflows to meet AI tools’ capabilities, instead of the tool adapting to fit our needs.
Discovering Qthena – Better results ‘out of the box’
Then we discovered Qthena, a generative AI solution from Questel. We achieved much better results using Qthena, right out of the proverbial box. Upon initial assessment, the solution demonstrated skills similar to a patent associate, including patent prosecution, drafting, and reasoning. Qthena produced quality work fast, requiring minimal time to produce meaningful results.
Testing Qthena – a thorough, collaborative process
First, we thoroughly tested Qthena using publicly-available information to benchmark the tool’s clear potential. Next, only with each client’s specific permission, we gauged whether it could perform safely and effectively with real client data. Data privacy, security, and confidentiality were crucially important. Even once a client was comfortable with our putting their data into Qthena, we had to verify security needs and data handling to uphold legal standards.
Qthena functioned like an associate attorney we could interact with who already knew about patents. The AI solution not only produced quality results quickly, but also our associates and partners remarked that it improved their writing quality. Our lawyers could ask Qthena to create a draft and provide strategic suggestions. The solution would generate a first draft, which our attorneys would review and update the results to quickly generate high-quality final products. Qthena could summarize art, quickly generate a structured draft, and accelerate our workflows. Encouraged by its performance, we decided to purchase Qthena for a group of IP attorneys.
Benefiting from Qthena, engaging in continuous dialogue
Several PLG attorneys are now using Qthena in various ways. Qthena does a good job of taking multiple different types of documents – whether it’s a writeup, a recording, a transcript of a recording with the inventors, etc. – to create an initial draft of a patent application or office action response. Qthena is a very good formal writer. We will often provide bullet points of the results we want, including tone or target audience, and Qthena does an excellent job of producing initial drafts.
In addition, we can have back-and-forth discussions with Qthena, asking questions and responding to answers, as we would with an associate. We can ask it to justify its rationale, cite documentary support for its conclusions, challenge an examiner’s interpretation, propose several different amendments or paths to go down, provide alternate examples in a patent application, and analyze different parts of each one. The productive dialogue we have with Qthena instills confidence that it understands requirements before doing the work. This kind of continuous engagement makes Qthena more accurate and thorough – it really increases my trust in the final result.
For example, when drafting patent applications, we can instruct Qthena to emphasize the concepts we identify as the most inventive, and it will appropriately focus and scope the draft application. For office action responses, Qthena provides more than just a draft – it also can propose a strategy based on the documents provided for review.
We also use Qthena for pulling in content from the U.S. Patent Office (PTO) and other workflows – we like having the single user interface. Using Qthena to incorporate PTO content more seamlessly is one of the reasons that we started using it as opposed to other legal AI providers that are not built for patent prosecution. Having the ability to interface with the US patent office and pull documents in automatically definitely saves valuable time.
The Qthena difference
Many AI tools we tried required a huge amount of effort up-front to learn to use them and customize prompts for their systems. Qthena is flexible, powerful and pre-trained with innate knowledge of patent workflows. As with most tools, it initially took time to understand, but now the solution easily fits into our IP practices. Our partners have found the right places to plug in Qthena – the solution has provided different benefits for everyone. Some people use it to stress-test their content, others to prepare initial drafts of applications and responses.
Our customer service experience with Qthena has been great and very responsive. We have access to high-level leadership within the company, people who really know the product and can get things done. Qthena made it possible for us to have servers hosted in the US which we needed to accommodate our biggest clients. The team accelerated the timeline for US hosting to accommodate our needs.
Safety belts and guardrails
Qthena has built-in guardrails and prompts which keep it grounded in the documents we provide, which gives us confidence in the results – this is critically important. The guardrails/safety nets help us to avoid hallucinations common to other AI products.
Qthena’s upfront discussions with the user verify that its understanding is correct, pointing to where it extracted information in the source documents. Qthena’s inherent structure categorizes the different types of input we provide and automatically instructs the underlying LLM to properly view our various input types. This connects draft claims to which parts of the specification it can do, compares arguments, and even helps to understand what the examiner is saying in an office action. Qthena’s ability to check that work is significant.
Other AI systems often quickly and confidently say, “Here’s the answer,” or “Here’s something that I found in this case file,” but it turns out either the case doesn’t exist or says something much different than what the AI represents. It’s frightening how superficial they are, and how quickly they provide a completely wrong answer.
With Qthena, our ability to understand the results, to know where those results came from, and to understand the accuracy of those results, is really more effective than any other system. The fact that Qthena has skills similar to a digital associate – built to analyze and confirm first, then write, and then provide support – has been one of the most important and meaningful impacts of using the system.
Closing thoughts
Qthena has been a great benefit to Potomac Law Group. We appreciate the way Qthena checks its understanding of the invention before it drafts a patent application or office action response, and engages in dialogue with the user. I haven’t seen other AI tools do this as effectively. We are planning to expand Qthena’s use at the firm within the IP team – there has been a lot of interest based on what has been accomplished so far.
Headquartered in DC and with lawyers in 25 states, PLG is one of the fastest growing firms in the country.
Legal IT Insider doesn’t charge for posts, which appear here based on merit. To submit an idea, contact [email protected]
The post Guest post: Potomac Law Group “transforms patent application with Qthena AI” appeared first on Legal IT Insider.