Lusha Chen (Dify) + Terry Purcell (TiDB) - live fireside, no slides, Nov 25 ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­  
When Your Users Are AI Agents

there,

When an agent fans out across parallel queries, holds state across hundreds of steps, and retries on failure without ever pausing, the data layer either keeps up or it becomes the ceiling. Most teams find out which one they have after the pilot ships. The assumptions that held for human traffic - predictable request patterns, modest concurrency, short-lived sessions - don't map cleanly to agent workloads, and the gap shows up fast.

 

On November 25, Lusha Chen, General Manager of APAC at Dify, and TiDB Field CTO Terry Purcell will work through this directly in a 30-minute fireside chat - no deck, no demos, just a real conversation about what breaks and how to build for what comes next. Dify's platform has more than a million applications deployed, which means Lusha's team is seeing these patterns before most.

 

Live Q&A at the end. One week away.

 

When Your Users Are AI Agents: What Breaks in the Data Layer


Date: November 25, 2026
Time: 11:00am CEST

Register for the Live Session

What you’ll learn

  • Which data-layer assumptions break first under agent traffic and why they break there
  • What the Dify team is seeing in production across 1M+ deployed AI applications
  • How memory, state, and bursty concurrency work differently for agent workloads
  • What builders are asking their infrastructure to do now that they were not asking a year ago
  • Where this is going over the next twelve months
  • The one thing to do this quarter if agents are coming for your stack

Can’t attend live? Register anyway and we’ll send the recording to everyone who signs up.

 

Yours in Distributed SQL,

TiDB, powered by PingCAP

demo_white

Get a Demo of TiDB

X
LinkedIn
YouTube

Join the Conversation

TiDB-Logo-Full

440 N Wolfe Rd, Sunnyvale, CA, 94085

Copyright © 2026 PingCAP Inc. All Rights Reserved.

Unsubscribe Manage Preferences