06 AI for R&D
Claude for R&D
Run patent and literature intelligence with Claude and Claude Code, with citations you can check and one monitoring workflow per team running by the end of the day.
This course is training, not legal advice. For an AI Act risk classification of your systems, see the consulting page.
- Duration
- 1 day
- Delivery
- On-site or online
- Group size
- Up to 12
- Languages
- Czech or English
- Prerequisites
- A Claude plan that includes Claude Code; we check the setup in advance
- Location
- Across Europe (in-house)
- Investment
- From €1,850 per in-house day (45,000 CZK); up to 12 participants, so from €155 per person
Who this is for
R&D and IP teams know what patents and papers are worth; the bottleneck is the hours it takes to search, read and summarise them. This day applies Claude and Claude Code to that work: search workflows, landscape summaries in which every claim links to its source, weak-signal monitoring, and data pulls from Espacenet, OpenAlex and arXiv. It is for R&D, innovation, IP and technology strategy teams. Technology Intelligence teaches the databases and the method; this course teaches running that method with Claude.
Key takeaways
- Run patent and literature searches with Claude without losing track of the query logic.
- Produce a landscape summary in which every statement points to a patent or paper you can open.
- Set up weak-signal monitoring on a technology field.
- Use Claude Code to pull data from Espacenet, OpenAlex and arXiv into a table you can analyse.
- Decide which material may go into an AI tool, and how to handle confidential R&D information.
- Leave with one monitoring workflow per team, already running.
What we cover
- 01
Claude on technical data
What Claude does well with patents and papers, where it invents or misreads, and how to set up work so every output can be checked.
- 02
Search and landscape workflows
Building and refining patent and literature queries with Claude; landscape summaries with citations that link to the source documents.
- 03
Weak signals and monitoring
Defining what to watch and what counts as a signal, then turning a one-off search into a recurring check.
- 04
Claude Code for data pulls
Scripts that query EPO Open Patent Services (Espacenet data), OpenAlex and arXiv, written and run with Claude Code. Participants do not need to be programmers: Claude Code writes the scripts and you review what they do.
- 05
Confidential material and build block
Rules for unpublished results and invention disclosures, then each team builds and starts one monitoring workflow on its own technology field.
Tools we use
How it's taught
Everything runs on your own technology field. Teams bring a topic they track, and the build block ends with a monitoring workflow that has already run once and produced output the team has checked against the sources.
About your lecturer
Jan Černý
Researcher & consultant, Competitive & Technology Intelligence
Jan Černý holds a Ph.D. in applied informatics (Prague University of Economics and Business) and a PhDr. in information science (Charles University). He has spent fifteen-plus years advising strategy, innovation and IP teams on competitive and technology intelligence, technology scouting and applied AI, and researches the same subjects academically, publishing in venues such as The Electronic Library, Review of Managerial Science, the Journal of Information Warfare and Online Information Review.
See research & publications- Ph.D., Applied Informatics, Prague University of Economics and Business
- PhDr., Information Science, Charles University
- 15+ years advising strategy, innovation and IP teams
- Peer-reviewed author in CI, OSINT, patent analytics and applied AI
- Delivered with the National Library and the Czech and Slovak Ministries of the Interior
Frequently asked questions
How is this different from the Technology Intelligence course?
Technology Intelligence teaches the databases and the method: scoping a landscape, reading CPC/IPC, citation analysis, scouting. Claude for R&D assumes that method and teaches running it with Claude and Claude Code. Teams new to patent work often take Technology Intelligence first.
Can Claude's summaries of patents and papers be trusted?
Only when they can be checked. A language model can misstate what a document says or cite a source that does not exist, so every workflow we build returns the source for each statement and includes a step where a person checks it. AI-driven patent-claim work shows real promise and predictable failure modes that have to be designed around (Černý, 'AI-Driven Patent Claim Construction', Business Information Systems / LNBIP, 2025).
Do participants need to code?
No. Claude Code writes and runs the scripts that pull data from Espacenet, OpenAlex and arXiv; participants describe what they need and review the result. Some comfort with installing software helps, and we check the setup before the day.
What about confidential R&D material?
We set the rules first: what may go into which Claude plan, what stays out (unpublished results, invention disclosures before filing, partner data under NDA), and how to work from public data and your own question rather than your own documents. Your company's policy takes precedence.
Which data sources do we use, and do they cost money?
Espacenet and the EPO's Open Patent Services, OpenAlex and arXiv. All three offer free access for this kind of use, some with registration or usage limits that we cover on the day. Claude usage depends on your plan.
Register your interest
Tell me which course fits and a bit about your team, and I'll reply within 48 hours with formats, dates and a tailored quote.
- jan@cerny.ai
- Phone
- +420 773 482 560
Typical response: within 48 hours.
Other workshops