Patent Bots®’s Post

We used our advanced machine learning software to review every patent issued – nearly 324,000! – by the USPTO in the past year. Nearly 89% of all patents contained at least one numbering, antecedent basis, or word support error. Such errors can increase the costs of obtaining a patent and lead to validity issues in the future. Fortunately, these errors are easily discovered and corrected using automated proofreading solutions like Patent Bots! #patent #patents #patentlaw #uspto #patentquality

Today, there are an impressive number of companies bringing AI and machine learning to patent drafting and prosecution. Use of AI/machine learning will become an integral part of patent prosecution. At a minimum, it will soon be an expected "best practice" to have AI software review patent applications prior to submission. Such reviews will provide an important quality check on the application. Further, as its capabilities increase and become more reliable, AI software will be a key part of each phase of the drafting process due to the ability of the software to propose edits/different approaches/alternative strategies.

Recalling that "Engineering Law" class I took during my Master's in Technology & Innovation Management in 2019.. It blew my mind! I realized that patents are a gold mine of creative solutions, just waiting to be unearthed. But here's the catch—navigating the world of patents can be a real headache, especially for individuals and small teams without a legal team or a hefty budget. So, I decided to build something to change that...and make something more 2024. I've been working on a GPT-powered tool that makes patent research faster, easier, and way more accessible for everyone. Think of it as your personal patent research assistant. 𝐓𝐡𝐢𝐬 𝐀𝐈-𝐩𝐨𝐰𝐞𝐫𝐞𝐝 𝐭𝐨𝐨𝐥 𝐜𝐚𝐧: 𝗗𝗶𝘃𝗲 𝗶𝗻𝘁𝗼 𝘁𝗵𝗲 𝗨𝗦𝗣𝗧𝗢'𝘀 𝗺𝗮𝘀𝘀𝗶𝘃𝗲 𝗱𝗮𝘁𝗮𝗯𝗮𝘀𝗲 𝗼𝗳 𝗼𝘃𝗲𝗿 𝟴 𝗺𝗶𝗹𝗹𝗶𝗼𝗻 𝗽𝗮𝘁𝗲𝗻𝘁𝘀: You can quickly find relevant patents using keywords, dates, or specific inventors. Say goodbye to those endless hours scrolling through pages of dense legal jargon! 𝗕𝗿𝗲𝗮𝗸 𝗱𝗼𝘄𝗻 𝘁𝗵𝗼𝘀𝗲 𝗰𝗼𝗺𝗽𝗹𝗶𝗰𝗮𝘁𝗲𝗱 𝗽𝗮𝘁𝗲𝗻𝘁 𝗱𝗼𝗰𝘂𝗺𝗲𝗻𝘁𝘀: This tool analyzes patent titles, abstracts, claims, and descriptions, summarizing the key innovations and their applications in plain English. 𝗦𝗽𝗮𝗿𝗸 𝘆𝗼𝘂𝗿 𝗰𝗿𝗲𝗮𝘁𝗶𝘃𝗶𝘁𝘆 𝗮𝗻𝗱 𝗳𝘂𝗲𝗹 𝗯𝗿𝗮𝗶𝗻𝘀𝘁𝗼𝗿𝗺𝗶𝗻𝗴: 𝗦𝘁𝘂𝗰𝗸 𝗶𝗻 𝗮𝗻 𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻 𝗿𝘂𝘁? This GPT can help you explore existing patents in your field and generate new ideas for products, processes, or even entire business models. 𝗡𝗲𝗲𝗱 𝗮𝗻 𝗲𝘅𝗮𝗺𝗽𝗹𝗲? Let's say you're working on a new design for a neurovascular stent or guidewires (one of my product specialties!). This tool can help you uncover patents related to existing stent designs, biocompatible materials, or drug-delivery mechanisms. You can then analyze those patents to identify potential improvements, avoid design flaws, or even discover entirely new approaches. 𝙋𝙧𝙚𝙩𝙩𝙮 𝙘𝙤𝙤𝙡, right? I'm always looking for ways to make this tool even better, so I'd love for you to give it a try and let me know what you think! 𝐘𝐨𝐮 𝐜𝐚𝐧 𝐚𝐜𝐜𝐞𝐬𝐬 𝐢𝐭 𝐡𝐞𝐫𝐞: https://lnkd.in/g8bAGTsC >ChatGPT... 𝗔+𝗣𝗮𝘁𝗲𝗻𝘁𝗧𝗵𝗶𝗻𝗸𝗧𝗮𝗻𝗸 Mode [𝗢𝗡]off #Innovation #PatentResearch #TechForAll #KnowledgeIsPower

Patent Bots® analyzed nearly 324,000 patents issued in the last year and found a bunch of scary deficiencies: Incorrect numbering, lack of antecedent basis, word support error. Here's some more scary patent errors. In Fiscal Year 2023, OPQA reported that 7.6 % of the allowances were not in compliance with at least one statutory requirement 20.7 % final rejections were not in compliance with at least one statutory requirement 21.7 % non-final rejection were not in compliance with at least one statutory requirement Overall, 16.5 % of all examiners' actions (allowances/rejections) were not in compliance with at least one statutory requirement #USPTO, Patent Bots®, #patent, #quality, #congress, #GAO, #OIG https://lnkd.in/gRVG3-Bj

Do you know what’s really scary? Numbering, antecedent basis, and word support errors in patents! 🕷️🎃👻 We used our advanced machine learning software to review every patent issued – nearly 324,000! – by the USPTO in the past year. Nearly 90% of all patents contained at least one numbering, antecedent basis, or word support error. Such errors can increase the costs of obtaining a patent and lead to validity issues in the future. But fear not... These errors are easily discovered and corrected using Patent Bots Prep & Pros Pro automated proofreading! #patent #patents #patentlaw #uspto #patentprosecution

What if you had a PhD sidekick for all your patent work? OpenAI just released their new o1 model. Insiders speak of PhD-level reasoning capabilities. Unlike its predecessors, o1 can "think" before responding. This approach is said to allow it to tackle complex, multi-step problems and self-verify its answers. What does this mean for the patent attorney profession? Some thoughts: (1) o1 is said to have unforeseen capabilities to reason about complex mathematical and technical concepts. Could such a model be a human-level sidekick for drafting highly complex patents like for advanced AI algorithms? (2) While previous models would only identify direct similarities in the data, an advanced model like o1 might see more nuanced connections. Could this be a leap forward in AI-assisted patentability, infringement and freedom-to-operate assessments? (3) By simulating human-like reasoning, might the model be able to predict patent examiner objections, allowing patent attorneys to address these proactively in the application? Just my initial ideas. What are your thoughts?

It's not April Fools'. In biting-the-hand-that-feeds news, an entity formed in 2021, Factor2 Multimedia Systems, sued the USPTO itself in the US Court of Federal Claims for patent infringement---note that an earlier-asserted patent (not asserted here) has already been invalidated under 101. Factor2's inventors have secured new counsel from that earlier litigation. This entity's connection to that earlier suit and entity is unclear. The earlier suit involved Bill Ramey, who appears to have been fired, and some connection to litigation funding---the inventors in the earlier case disclosed that others have a financial interest in the outcome of the case, but the district court there denied additional discovery into who or what the connection was; history and counsel suggest it was AIPI Solutions, and there appears to have been a connection to Delphinus Technology, Inc., but we may never know. Cases 1:24-cv-00475, 1:24-cv-00477 against the USPTO on "the system which provides two factor authenticated access to: USPTO.gov; MyUSPTO; MyUSPTO account; Patent Electronic System; EFS-Web; PAIR; and Patent Center, including the ‘2stable’ app[,] and distributes content and authenticates users on the USPTO site." Patents are 8,281,129; 9,703,938; 9,727,864; 9,870,453; 10,083,285; 10,769,297; previously invalidated patent (8,266,432), invalidated under 101 in ED Va.  Note, I edited this post after new information came to light, though to be clear, I, like most involved, are working off of imperfect information about AiPi and what their level of involvement in a lot of cases is or was. Unfortunately, the previous judge denied calls for disclosure.

🤔 Need to iterate on your #patent specification? 🪄✨ ”Ask AI” allows you to underline any sentence or paragraph and reformulate it with a basic #prompt. Improve your patent drafting to the next level, thanks to davinci! 🟣 Want to try it on your own? Join the waitlist in one click 👇😉

One key area of interest for me is how LLMs an related models will change patent prosecution and potentially increase patent quality. Chu, Lo, Hsiang and Cho in their paper from PARIS to LE-PARIS (https://lnkd.in/gz4zd4Aw) demonstrate the efficacy of LLMs for automating responses to Office Actions by combining a qualitative evaluation of the responses as well as assessing practical utility of the systems. The validation focused exclusively on responses to USPTO Office Actions, it would be interesting to understand the model's efficacy for responses to non-USPTO office actions.

Unlock the World of Intellectual Property Rights! We're excited to announce the 8th Edition of the Patent School Workshop, hosted by CIAP in collaboration with the Department of Computing Technologies and our esteemed knowledge partner, Lexorbis. Mark your calendars for October 15th-17th, 2024! Whether you're a student, professional, or entrepreneur, this workshop is your gateway to understanding the intricacies of Intellectual Property Rights (IPR) and boosting your career in this ever-evolving field. What You’ll Learn: - Introduction to IPR - Patent Search and Drafting - Prosecution Process - Patent Landscape Analysis - Technology Evaluation for Patentability Why Attend? Gain practical insights, network with industry experts, and enhance your skills. This is more than a workshop – it's an opportunity to elevate your professional journey! Registration Fee: ₹200 (inclusive of taxes). Don’t miss this chance to empower your understanding of IPR. We look forward to seeing you there! #IntellectualProperty #IPR #PatentWorkshop #ProfessionalDevelopment #Lexorbis #CareerGrowth