IBM TechXchange 2026
Docling sessions at IBM TechXchange 2026 in Atlanta, October 26–29
The Docling team is at IBM TechXchange 2026 with talks and hands-on labs. In the labs you build something with Docling for IBM watsonx instead of watching a demo, and you can keep going after the session with a free trial.
Rooms and times are taken from the session catalog. Check it for last-minute changes and to add sessions to your agenda. All times are EDT.
Monday, October 26
Agentic RAG with OpenRAG on IBM watsonx.data & External Agents: From Docling Ingestion to MCP
Thomas Schäck (IBM), Charna Parkey (IBM)
An agentic RAG pipeline with OpenRAG on IBM watsonx.data: Docling parses the documents into an OpenSearch knowledge base, and agents query it over MCP from watsonx Orchestrate and the Hermes Agent framework. Run by the watsonx.data team.
Tuesday, October 27
Enriching the Knowledge Layer for Enterprise Agents with IBM and NVIDIA
Ryan Chesler (NVIDIA)
A case study on turning complex enterprise documents into agent-ready knowledge with Docling, DocLang and NVIDIA NeMo Retriever, while keeping layout, metadata and provenance intact on the way to RAG and agents.
Build a RAG Pipeline and Document Agent with IBM Docling, Amazon Bedrock, and Amazon S3 Vectors
Michele Dolfi (IBM)
A citation-aware RAG pipeline on IBM and AWS: Docling parses the documents, and Amazon Bedrock and S3 Vectors handle embeddings, retrieval and generation. Then a Strands agent navigates documents by their outline and uses the RAG pipeline as one of its tools.
Wednesday, October 28
From Bucket to RAG: Event-Driven Docling for watsonx and Confluent
Michele Dolfi (IBM), Peter Staar (IBM)
Drop a file into a bucket and watch it flow: Docling for IBM watsonx produces layout-aware chunks straight onto Kafka, and Apache Flink on Confluent gates quality, removes duplicates, redacts PII, embeds with watsonx.ai and indexes into OpenSearch. No scheduler, no glue script.
Docling for IBM watsonx: Convert, Extract, and Build Your Own Document-Powered App
Michele Dolfi (IBM), Peter Staar (IBM)
Convert documents in the Workbench, then call the same service from Python. Then pick a track: a RAG notebook with Docling chunking, Docling in agentic frameworks, or an end-to-end app built with IBM Bob.
Docling for IBM watsonx: Building Agentic Document Workflows at Enterprise Speed
Michele Dolfi (IBM), Eduardo Monich Fronza (AWS), Mikhail Shapirov (AWS)
Context engineering for agents: why plain PDF-to-text loses the headings, tables and reading order agents need, and how Docling for IBM watsonx delivers structured, agent-ready output as a managed service.
The Missing Layer in Enterprise AI: Building Document AI Pipelines with DocLang
Peter Staar (IBM), Max Vermeir (ABBYY)
Why documents break AI pipelines, and how DocLang, the open AI-native document format created by ABBYY, IBM, NVIDIA and Red Hat, makes them reliable. With Docling and ABBYY FineReader Engine examples and benchmarks.
TEC-2232 and TEC-2541 run at the same time, in neighbouring rooms in Building B.
Before and after the event
- Start now. The labs run on a free trial of Docling for IBM watsonx.
- Keep going. The guides and connectors pick up where the labs stop.
- Meet us. Come to a session or find us between them. We are happy to talk about your use case.