TiDB SCaiLE Europe 2026 Agenda: Why You Should Attend 4 June

Key Takeaways

When and where: Thursday, 4 June 2026, 09:00–19:30 at Epicenter Stockholm. Registration is free.

Who’s speaking: Bolt, LHV Bank, AWS, Zilliz, Ververica, and more alongside TiDB engineering, all on a single track.

What you’ll get: Practical depth on MySQL migration, native vector search, agent memory, and streaming decisions for agentic AI.

The keynote: TiDB Co-Founder and CTO Ed Huang on why the next decade of computing won’t be a refinement of the last.

Most teams shipping AI agents in 2026 hit the same wall around the same time. The prototype works. Ten users or even a thousand users mostly work. But then one user action triggers thousands of agent instances, context has to branch per agent in milliseconds, and vector lookups stack on top of transactional reads. The multi-database stack holding it all together starts paying for itself in reliability incidents and engineering overhead. That is the failure mode the TiDB SCaiLE Europe 2026 agenda is designed to solve.

The event runs Thursday, 4 June 2026, 09:00 to 19:30 at Epicenter Stockholm, under the theme Scaling Data. Accelerating AI. Enabling You. It is a full-day, single-track gathering for engineers, architects, and platform leads running agentic AI workloads in production, or close to it. If you are evaluating how to put agent memory, vector retrieval, and transactional state into a single system that survives at scale, this is where the people who have already done it will be.

Rethinking the Database Layer for Agentic AI

Every previous wave of application scaling assumed humans on the other end. Agentic AI breaks that assumption. A single user action now spawns thousands of agent instances, each branching context, writing memory, calling tools, and reading shared state in real time. Read-to-write ratios collapse. Concurrent session counts go up by orders of magnitude. The bottleneck moves from the application tier to the data layer.

TiDB SCaiLE Europe makes the argument that fragile multi-database stacks (one system for transactions, one for analytics, one for vectors, glued together with ETL) cannot keep up. The answer? Converge those workloads into a single distributed SQL engine. That argument runs through Ed Huang’s opening keynote and every session that follows.

Sessions Built on Production Experience

List of speakers that make up the TiDB SCaiLE Europe 2026 agenda.

Every speaker on the TiDB SCaiLE Europe 2026 agenda either runs TiDB in production at scale or builds and deploys the technology directly. A few of the talks on the agenda:

Ed Huang, CTO of TiDB, opens with Welcome to the Age of Autonomous Systems, a keynote on why today’s cloud and infrastructure stack was never built for software where agents are the primary users, and what the next decade of computing actually requires.
Leandro Morgado, Senior Database Reliability Engineer at Bolt, on Breaking Up With MySQL: How Bolt Rebuilt for 100 Million Users on TiDB. Bolt grew 400% post-pandemic, hit the point where adding a single column to a 1 TB table took a week, and migrated their most critical workloads to seven TiDB clusters on AWS. This session covers the evaluation, the migration, and how they got to 3:1 compression and five-nines reliability in production.
Tarmo Kople, IT Infrastructure Architect at LHV Bank (and formerly Bolt’s DB team lead), on TiDB in Practice: Scaling Databases Without Losing Sleep. What deploying TiDB actually looks like across two very different organizations, from high-growth startup to regulated banking.
Bernard Kavanagh, Principal Solutions Architect at TiDB, on The Memory Wall: Why AI Agents Fail (and How to Fix Them). The architecture pattern behind millions of concurrent agents with persistent memory and consistent state, and the Decide-Validate-Remember loop that production teams use to reduce token cost and avoid state drift.

Each session is structured to leave you with something specific: A pattern, tradeoff, number, or a config decision you can take back to your own architecture review.

Deep Dives on Vector, Migration, and Streaming

The technical track goes past the keynote pitch into the architecture questions practitioners actually run into:

Mattias Jonsson, Principal Software Engineer at TiDB, on Vector Search Meets Distributed SQL: Why Agentic AI Doesn’t Need Another Database. How vector embeddings, semantic search, and agent memory patterns work alongside transactional and analytical workloads in a single query layer, and what that means for the stack you build.
Daniël van Eeden, Technical Solutions Engineer at TiDB, on Migrating From MySQL to Distributed SQL: What Changes, What Doesn’t, and What Breaks. The queries that run identically, the assumptions that quietly break, and the Raft and LSM tree internals that explain why. Drawn from real production migrations across Europe.
Simon Hearne, Founding Solutions Architect at Zilliz, on When EXPLAIN Isn’t Enough: Visualising Vector Search for Engineering and Product Teams. What embeddings look like in high-dimensional space, what quantization does to recall, and how to catch retrieval failures before your agents do.
Fredrik Tormod and Anja Pettersson, Solutions Architects at AWS, on AgentOps Meets SaaS. The underexplored third dimension of agentic patterns: Agents running the operations of a SaaS, not just the product.
Alex Campos, Technical Sales Professional at Ververica, on Agentic Streaming Decisions: Unlocking Real-Time AI With Ververica and TiDB. How Apache Flink and TiDB combine to give agents the fresh data, durable memory, and trusted context they need to act on business events as they happen.
Muhammad Ahsan Ayaz, Software Architect at Scania Group, on Zero to Agentic Orchestra With Google ADK. A live-coding session building a multi-agent system in Python with Google’s ADK and MCP servers, covering three orchestration patterns (sequential, parallel, iterative). ADK gives you control flow, MCP gives each agent its capabilities, and you leave with working code.

The day closes with a panel and fireside chat, then two hours of networking and happy hour from 17:30 to 19:30.

Why Stockholm, Why 4 June

Europe’s AI infrastructure community has historically had to fly to the US to get this kind of single-track, practitioner-led depth. TiDB SCaiLE Europe brings it within a short flight of London, Berlin, Amsterdam, Paris, Helsinki, and Copenhagen. Epicenter Stockholm is a working tech venue in the center of the city, not a hotel ballroom. The day is structured so European, UK, and Middle Eastern attendees can travel in and out without losing a week.

Who Should Be in the Room

The TiDB SCaiLE Europe 2026 agenda is built for:

Developers and engineers building applications on distributed or cloud-native databases.
Data and platform architects designing infrastructure for scale and resilience.
Engineering leaders evaluating modern data strategies for AI-native workloads.
Technical decision-makers exploring alternatives to legacy database infrastructure.

If your job description involves the words “agent,” “memory,” “vector,” “multi-tenant,” or “scale,” this agenda is worth the trip.

What You Will Leave With

Takeaway	How You Use It
Production patterns	Concrete architectures from Bolt, LHV Bank, AWS, Zilliz, and Ververica you can map to your own stack.
A sharper view of the data layer	A clearer read on which database primitives actually matter for agentic workloads.
Migration and vector specifics	Named tradeoffs from teams who have already done MySQL-to-distributed-SQL migrations and shipped production vector search.
Peer connections	A single-track room and a two-hour happy hour means the hallway conversations are just as valuable as the sessions.
Roadmap visibility	First look at where TiDB and the surrounding AI infrastructure landscape are heading next.