ConVEx San Jose - Gen AI and iiRDS

Looking back at ConVEx 2025 in San Jose, generative AI was still everywhere. While some are still trying to figure out its impact on technical writing, others are stressing that everyone needs to incorporate it into their daily work. Here are some paraphrased quotes: 

"Every team needs a person dedicated to generative AI."

"Gen AI won’t take your job, it's the tech writers who know how to use it."

Reliable AI 

Gen AI is certainly not without its flaws. You need to know how to use it. So it's no surprise that the reliability of gen AI is back on the agenda. How to deal with the hallucination problems of large language models (LLMs) remains a hot topic. 

One answer is Retrieval Augmented Generation (RAG). RAG adds information from a curated content pool to the context window of the LLM. However, there are different ways of retrieving information. First, there are approaches based on vector search. Then there's the new Model Context Protocol (MCP). MCP allows a standardized extension of the closed data pool of an LLM. This will make it easier to integrate other data pools and use the data with LLMs. MCP may be on the verge of enabling the new RAG 2.0. 

Whatever future developments bring, graph-based RAG has proven to be an effective way to control your chatbot. No wonder it was mentioned in several talks. And of course, with graph-based semantic RAG, the importance of metadata was highlighted. 

iiRDS and metadata 

When there’s metadata, there’s the intelligent information Request and Delivery Standard iiRDS. On its way to becoming an IEC standard (IEC PAS 63485:2023), iiRDS is gaining traction internationally. There were several talks about iiRDS at ConVEx. 

iiRDS specifies an exchange format for technical documentation that adds a graph-based metadata layer to your content. Using the W3C standard RDF, you can exchange content of any format with an accompanying metadata side-file. This is all done in a standardized way. 

In my talk on business opportunities and iiRDS, I showed how current projects are using metadata. Metadata should not be lost during publication but is an additional artefact that is published with your content to improve the user experience and to make your content AI-ready. 

After the talks and between the sessions, there were many interesting discussions at the tekom and iiRDS booth. That’s what’s really makes these events shine, the tech doc community coming together to discuss the latest trends and best practices. 

Thanks to everyone who participated in the discussions and visited the booth. Thanks also to CIDM for organizing this great conference. It was a pleasure to attend ConVEx San Jose.