Newsletter
GenAI & IP
Issues with use cases, prompt examples and fundamentals on generative AI in patent practice.
Welcome
Newsletter area
Individual issues may be released publicly; protected issues require the newsletter password.
Latest
Latest issue
A walk through Anthropic's new Legal Plugin
A walk through Anthropic's new IP plugin: twelve skills, one scheduled agent, five MCP servers. Where the architecture is instructive, where the US content breaks at the EPC border, and what it means for your own practice.
Archive
Previous issues
Why agents are so reliable, and how I replicate that in the browser
Anyone using agent-based AI tools like Claude Code or Cursor notices a kind of reliability the regular browser chat often lacks. The reason sits in how these tools manage context. Pillar 1 of 4 in our prompt-engineering-for-patent-practice series.
Gemma 4: a first take from patent drafting
Google has released Gemma 4. In my drafting tests it was twice as fast as Gemma 3 and produced better quality. Who should switch, and who is better off staying with a smaller model.
Closing the Circle — Part 2
Practical AI prompts for patent attorneys implementing hybrid workflows — local models for confidential data, cloud AI for published information.
Closing the Circle — Part 1
The cloud vs. local dilemma — information bleeding, safe implementation strategies, hybrid workflows.
Closing the Circle
Confidential cloud prompting, local AI, and the epi AI guidelines for European patent attorneys.
Patent Portfolio Analysis
How LLMs like Claude generate and execute Python code to analyse large patent datasets effectively.
Knowledge Graphs aus Patentdaten
Structured data extraction, conceptual mapping, and HTML-based visualisation of patent relationships.
Patent Data Analysis — Part 2
Data preparation with Python-based data mining; accessing the EPO Technology Intelligence Platform (TIP).
Patent Data Analysis, Landscaping
Using LLMs for patent data analysis without specialised software — six analytical approaches and a three-step implementation.
Fundamentals: GenAI Overview
The dual nature of generative AI — analytical strengths and generative applications for patent work.
AI Overview
An overview of the AI landscape: AI, machine learning, neural networks, deep learning, GenAI.
Flow Chart Generation
Automatically create a flow chart for a process claim — using Graphviz and an LLM as code generator.
Pitfalls Of Using Generative AI
Why LLMs are often overrated — hallucinations, statistical not cognitive systems, realistic limits.
Prompt Engineering Basics
One-shot and few-shot learning, cues for output control, and five best practices for precise prompts.
Patent Language Style Guide Enforcement
Automated style-guide and compliance checks for patent applications.
Consistency Check
Automatically check the terminology consistency of a patent application draft.
How to use generative AI for automation
First orientation — "Don't use AI when there's an easier option." Which tasks are worth automating.
Willkommen — GenAI & IP Newsletter
Welcome issue: a year-long introductory course on generative AI for intellectual property — use cases, tips, fundamentals.
Not yet a subscriber? Sign up on boesherzgoebel.de.