Search Solutions Tutorial Day
Come to join our Search Solutions 2026 Tutorial day!
Synopsis
Join us for the Search Solutions 2026 Tutorial Day, organised by the BCS Information Retrieval Specialist Group (IRSG), featuring two timely tutorials at the intersection of Search and AI.
Tutorial 1: The Frontiers of Open-Source Agentic Search
Alessandro Benedetti, Sease Ltd. & Radu Gheorghe, Vespa AI
Explore the emerging world of agentic search, where AI agents move beyond traditional query-response systems to plan retrieval strategies, interact with tools and synthesise answers. This practical tutorial covers open-source technologies, the Model Context Protocol (MCP), evaluation, security and agentic capabilities across Apache Solr, Elasticsearch, OpenSearch and Vespa, including hands-on demonstrations.
Programme:
Part 1: Foundations & Paradigms (90 Minutes)
- 00:00 - 00:30 | Foundations of Agentic Search (30 mins)
- 00:30 - 01:00 | The Open-Source Agent Orchestration Ecosystem (30 mins)
- 01:00 - 01:30 | Evaluation Metrics & Benchmarking (30 mins)
01:30 - 01:45 | Morning Break (15 mins)
Part 2: The Lucene Ecosystem Deep Dive (90 Mins)
- 01:45 - 02:05 | Elasticsearch Support for Agents (20 mins)
- 02:05 - 02:35 | OpenSearch Native Agentic Framework (30 mins)
- 02:35 - 03:15 | Apache Solr as a First-Class Agent Citizen (40 mins)
03:15 - 04:15 | Lunch Break (60 mins)
Part 3: Hands-On Solr & Vespa Deep Dive (135 Mins)
- 04:15 - 04:45 | Hands-On: Running Agentic Search with Apache Solr (30 mins)
- 04:45 - 05:30 | Vespa Platform Architecture for Agents (45 mins)
- 05:30 - 06:30 | Hands-On: Running Agentic Search with Vespa (60 mins)
06:30 - 06:45 | Afternoon Break (15 mins)
Part 4: Security, Guardrails & Future Frontiers (45 Mins)
- 06:45 - 07:15 | Security Boundaries & Agent Guardrails (30 mins)
- 07:15 - 07:30 | Wrap-up, Q&A, and Research Frontiers (15 mins)
Tutorial 2: Inside Language Models: From Probing to Mechanistic Interpretability
Kevin Roitero & Riccardo Lunardi, University of Udine
Discover what happens inside the LLMs powering modern search and RAG systems. The tutorial introduces probing and mechanistic interpretability, covering hidden representations, transformer circuits, activation patching, sparse autoencoders and feature steering, complemented by practical demonstrations.
Designed for search and IR practitioners, researchers, developers, data scientists and AI professionals, the Tutorial Day offers an excellent opportunity to gain practical insights into both building intelligent search systems and understanding how they work.
Programme:
- 00:00–00:15 Motivation: understanding modern LLMs and RAG beyond black-box.
- 00:15–01:00 Probing: representations, hidden states, probe design, layer-wise analysis, baselines, control tasks, interpretation pitfalls.
- 01:00–01:15 Practicum I: probing hidden representations in LLMs, including layer-wise probing, probe training, visualization, and interpretation of results.
01:15–01:25 Coffee Break.
- 01:25–02:10 Mechanistic interpretability: transformer circuits, residual streams, circuit analysis, activation patching, and feature steering.
- 02:10–02:25 Practicum II: activation patching, steering vectors, and causal analysis of model behaviour using transformer models.
- 02:25–02:40 Modern directions and practitioner use cases: sparse autoencoders, feature steering, and open challenges.
- 02:40–03:00 Discussion, Q&A, and guidance for applying interpretability techniques. Detailed Description
The tutorials take place on 24 November 2026, ahead of the Search Solutions Conference on 25 November.
Good to know
Highlights
- 9 hours
- In-person
Refund Policy
Location
BCS, The Chartered Institute for IT
Ground Floor
25 Copthall Avenue London EC2R 7BP
How do you want to get there?
