Payment infra for humans and AI agents: Clink on how data is changing when agents become first-class participants in commerce - and why one flexible schema on TiDB is what holds it together.
Watch video: link

Payment infra for humans and AI agents: Clink on how data is changing when agents become first-class participants in commerce - and why one flexible schema on TiDB is what holds it together.
Watch video: link
"We need a data solution where it rollbacks." โ Julie Wang, Co-Founder & CMO, Clink. In agent commerce, every payment has to complete fully or not at all. TiDB made that possible at scale. ๐ link
When your users are AI agents, what breaks in the data layer? Lusha Chen (Dify) + Terry Purcell (TiDB) on what changes when the user is a machine. Oct 29, 11am CEST.
Register: link
What makes a database AI-native? Lusha at Dify: one system for everything - no fragmented tools. For Dify's millions of AI workflows, that's TiDB.
Most AI agents aren't stateful โ they're prompt chains with amnesia. Agent state is what actually lets them recover from crashes, resume multi-step tasks, and audit what happened.
Clink needed one schema to handle every payment method โ for humans and AI agents alike. Co-founder and CMO Julie Wang on why TiDB made that possible: link
Your AI agent has no memory. It has storage. Ed Huang on why persistence isn't enough - selection, compression, decay, and contamination prevention are where agents actually break: link
AI can write your code. Reviewing it for production is still your job - especially when it's going into TiDB.
Guide: link
This Thursday โ live session on architecting databases for agentic AI.
Sep 24 at 11am CEST. Register: link
This Thursday โ join us live at 11am SGT for a session on building data infrastructure for AI agents.
Sep 24. Register: link
Last call, Montevideo!
Percona University lands this Wednesday, Sept. 23, for a free day of #PostgreSQL, #MySQL, #TiDB, observability, AI and more.
Register: https://bit.ly/4y0NOm5
#PerconaUniversity #OpenSource #Databases #Uruguay #AI #Observability #DatabaseManagement
Kimi K2.6 builds full-stack websites from a text prompt. Ed Huang breaks down why per-tenant Postgres failed and what the architecture that actually works looks like: link
Sep 24 at 11am CEST - architecting databases for agentic AI, with lessons from production at massive scale.
Register: link
APAC: Join us Sep 24 at 11am SGT - live session on building the data layer that keeps up with AI agents at scale. Register: link #AgenticAI #TiDB
Novita gives AI agents a live database with one parameter: database=True
Zero ops. Instant provisioning. Fully isolated.
How they built it on TiDB:
Serverless database โ managed database.
One asks you to pick an instance size. The other decides capacity itself, scales to zero, and bills for usage.
That gap matters for AI workloads. Brian Foster explains the tradeoffs:
SaaS scale โ agentic AI scale. Same consolidation problem, higher frequency.
Terry Purcell (TiDB Field CTO) on the architectural arc and what builders need to do now.
Sep 24 ยท 11am CEST
Register: link
SaaS scale โ agentic AI scale. Same consolidation problem, higher frequency.
Terry Purcell (TiDB Field CTO) on the architectural arc and what builders need to do now.
Sep 24 ยท 11am CEST
Register: link