University of Portsmouth pre-launch startup makes patent research easier

Helping startups save time with better patent insights

After establishing an advice clinic for student and graduate entrepreneurs at the University of Portsmouth called iPass (Intellectual Property Advice Support Service), Joe realised the immense challenge faced by fledgling innovators in locating the patent data needed to guide their startups. Most patent databases are notoriously complex, requiring substantial expertise to navigate and interpret. Recognising the need for a more user-friendly solution, Joe was inspired by the burgeoning field of machine learning and AI tools and initiated a dialogue with Fredrik about developing a search tool that would simplify patent research for startups.

One significant hurdle with patent databases is that users struggle to optimise searches for existing patents that might be similar to their own ideas. This critical knowledge could enable them to change direction and refine their concepts, potentially saving months or even years of wasted effort on ideas that might have led to a dead end.

Additionally, larger organisations hold an advantage over smaller startups that can't afford to hire patent lawyers, making it challenging for these smaller entities to bring their ideas to fruition unless acquired. Joe and Fredrik believed that IPitome could help to balance the patent landscape. SoftwareOne began building the IPitome tool to help achieve the founders’ goals.

Building an effective, AI-powered patent search and results solution

The SoftwareOne team worked with Joe and Fredrik to build a solution that could deliver the capabilities they were looking for. The IPitome platform uses Amazon Kendra, an intelligent enterprise search service, to help users find the information they need.

The solution will also use AWS Amplify to deploy the custom application as a web and mobile app. The IPitome platform can search through 100,000+ patent documents stored in Amazon S3 storage and in metadata to quickly identify matches to queries from the user interface.

Throughout the development, SoftwareOne’s certified AWS Solution Architects provided guidance on best practices while implementing and optimising the platform.

Developing the solution required substantial testing to refine the model, with results verified by checking against the results from standard patent database search tools. To limit the amount of data that had to be managed, all the tests involved searching a subset of just one patent database. The SoftwareOne team met regularly with Joe and Fredrik to discuss how the platform was performing and where it could be further optimised. Collaborating with SoftwareOne, Joe focused on developing the user interface and traffic light search results, while Fredrik concentrated on helping to develop the backend functions.

The SoftwareOne team designed the platform with a simple user interface and included a traffic light system to help users determine the patentability of their ideas. After entering keywords into a search box, users see results categorised into three areas: red for ideas already covered by existing patents with legally enforceable rights, amber for ideas that warrant further investigation to determine whether they’re patentable and green for ideas that appear to offer new patent opportunities. This makes it easier for users to understand where they can find the most relevant patent data before they begin making any commercial plans.