Are You Ready for the Next Evolutionary Step in Artificial Intelligence?

The IP industry has undergone a breathtaking transformation in recent years. Artificial Intelligence has rewritten the rules of the game when dealing with patent information. Work that was difficult before AI is now done in the blink of an eye thanks to the use of new methods and tools. More and more patent department processes are being enhanced with Artificial Intelligence in order to save resources and optimize process efficiency. predori has been a key driver of this development since 2020. Many of the functions that people have already become accustomed to and no longer want to work without were originally developed by predori. One example of this is the “Ask your Data” feature, which was already available in the predori application at the beginning of 2023. 

As an AI-first company, we always remain true to our credo: AI must be comprehensible and transparent. This is the biggest challenge, because this characteristic is indispensable in the IP industry in particular. From the very beginning, we have developed predori in such a way that the AI is a helpful assistant in interaction with humans and the work of the AI is always documented. 

It is clear that AI is becoming more and more deeply integrated into processes in the IP industry and that more and more IP experts are using modern technologies. The big task for us was therefore to incorporate super intelligent autonomous AI systems into our product portfolio while still living up to our credo of traceability and transparency.  

We would like to discuss with you how predori is introducing the next evolutionary stage of Artificial Intelligence in the IP sector so that users can benefit from even greater efficiency, using two use cases from the everyday life of a patent department. The first use case concerns patent monitoring, i.e. the monitoring of technologies and competitors. The second case concerns the world's first autonomous patent search system, which we will be presenting for the first time at IP-Service World 2024. 

Patent monitoring creates the basis for a variety of essential tasks of a patent department. Be it informing the development department about current developments of competitors, avoiding patent infringements or reporting developments in entire fields of technology to the management. To fulfill these tasks, it is essential to get the right documents to the right recipients in the right way. This is where the art lies: every patent department knows the three big problems when it comes to patent monitoring: 

1. By-Catch 
   There is no such thing as the perfect monitoring profile. You will always have by-catch, i.e. documents that match the search-string but are irrelevant in terms of content. 
2. Accessibility of Knowledge 
   Monitoring is an internal company service. As a patent department, you have to prepare the data according to its users. However, understanding patent literature and accessing knowledge is a major problem for the vast majority of engineers. 
3. Structuring Knowledge 
   Monitoring is not used for short-term information, but for the long-term development of structured knowledge that can be accessed in a targeted manner. However, large amounts of data and the complexity of patent literature make it very difficult to assign knowledge. 

At predori, we investigated different approaches. These included pre-trained classifiers, trainable classifiers, rule-based systems and an Artificial Intelligence Agent System. Classifiers were ruled out for practical reasons alone, as it turned out that the training effort required to achieve an acceptable quality of results was far too great. Quite apart from this, the training status can only be judged on the basis of the result; how the system works, what it considers relevant and why, is completely hidden. Rule-based systems have proven to be surprisingly resolute. In addition, rule-based systems fulfill our basic requirement of traceability due to the transparently comprehensible rules. However, rule-based systems have proven to be too unwieldy for daily use in patent departments.  

The agent-based AI systems developed by predori outperformed all other systems examined in terms of quality. At the same time, we have succeeded in structuring agent-based AI systems architecturally in such a way that they are transparent and comprehensible, which makes them ideal for use in patent departments. 
 

predori relies on a specially developed AI agent system for monitoring that outperforms previous classifiers and is also traceable. 

By using AI agent systems, we can offer our customers the following added value in monitoring: 

• AI Auto-Filtering 
  Effectively filters out by-catch. This saves you valuable time and increases the quality of your stored data. 
• AI Auto-Summary 
  Automatic summarization of documents. This means you have to read less. 
• AI Auto-Answering 
  Automatic answering of predefined questions. With this function, you can work proactively with the content of documents and save yourself tedious reviews.  
• AI Auto-Tagging 
  Automatic structuring of knowledge with individually predefined tags. This neatly structures your internal knowledge collection, making it easier to access knowledge. 

We would be happy to demonstrate our AI-supported monitoring workflow to you in a personal live demo: book a demo 

What follows now admittedly sounds a bit like science fiction. Imagine you give an AI assistant the task of carrying out a patent search. After a short training period, the system asks the necessary questions, which you can answer in dialog. Now the AI assistant gets to work and carries out the search together with a whole orchestra of specialized AI agents.  

Science fiction? Not quite, predori is already capable of conducting novelty searches in mechanical engineering, electrical engineering and physics in exactly the same way. The advantages are obvious: 

• Time: You save valuable time because you don't have to search or evaluate. What you get are results on the basis of which you can continue working immediately. 
• Money: The research capacities are virtually inexhaustible. 
• Resources: Very few companies have as many patent searchers as they need to answer all the questions that arise. At the same time, very few engineers, innovation managers and product managers are really good patent searchers. Apart from that, finding and qualifying good patent searchers is not easy and involves high costs. 

According to previous comparative tests, the quality of the results is comparable to human research.  

There is one aspect in particular where we at predori remain true to ourselves: the results of the AI agent search are completely comprehensible. The AI explains exactly which search strategy it followed, which classes were considered, which terms were searched for and which criteria were used to evaluate content. predori's autonomous AI search is based on an AI agent system whose agents have been trained and qualified to carry out searches. 

Have you ever heard of agent-based AI systems? Simply explained, an AI agent system is the combination of several specialized AI agents that divide complex tasks among themselves so that subtasks are each processed by the agent that is specialized and trained for the task. Together, an AI agent system can therefore also solve more complex tasks. 
 

With its autonomous patent search, predori is ushering in the next evolutionary stage of Artificial Intelligence in the IP sector. 
 

We as humans give the AI agent system a task, whereupon the system independently determines the optimal solution path and then solves the task along this solution path. The next evolutionary step in Artificial Intelligence is therefore for Artificial Intelligence to be operated by Artificial Intelligence. The special feature of predori's AI agent system is that every step of the solution path can be traced, making the system completely transparent. Incidentally, predori has several patent applications for this. 
 

In predori's unique AI agent system, AI works with AI to solve complex tasks. As humans, we benefit from the high quality and explainability of the results. 
 

Be there when predori reports publicly for the first time on the next evolutionary stage of Artificial Intelligence at IP-Service-World 2024:  

Monday, 25.11.2024 at 11:55 o'clock 

Link: https://www.ipserviceworld.com/speaker/stefan-brehm/ 

AI agent systems are orchestrated and coordinated systems that can perform complex tasks such as patent searches and patent monitoring autonomously or semi-autonomously. Key aspects of predori's AI agent system are the traceability of results, the ability to engage in dialog in order to make improvements together with users and the high quality of the results. predori is a global pioneer in this technology and is already using AI agent systems to the greatest satisfaction. 

Do you want to optimize your organization's efficiency in patent monitoring and patent searching? Arrange your personal live presentation: book a demo.