SINGAPORE, June 11, 2026 /PRNewswire/ — TiDB, the open-source distributed SQL database, today launched the TiDB Agent State Stack at SuperAI Summit Singapore, a unified data foundation that gives AI agents durable memory, persistent state, and continuous context as they scale into production.
As organizations move beyond AI pilots and proofs of concept, a new challenge is emerging: AI agents struggle to retain memory, maintain state, and preserve context across sessions, workflows, tools, and teams. While many agents can perform tasks in isolated environments, scaling them into enterprise-grade systems often requires stitching together multiple disconnected components for memory, retrieval, file storage, and operational data.
The pressure to solve this is intensifying. According to Gartner, fewer than 5% of enterprise applications included task-specific AI agents in 2025, a figure expected to reach 40% by the end of 2026. Yet while 75% of organizations are piloting or deploying AI agents, only 15% are working with fully autonomous agents, a gap that points directly to the missing infrastructure layer beneath them.
The TiDB Agent State Stack closes that gap. It brings memory, state, and context management onto a single distributed SQL foundation, so the data an agent needs to act reliably lives in one place rather than scattered across systems that were never designed to work together.
“The agents our customers run in production spin up and discard millions of database instances a day. That starts with the database: Agents need a backbone that is ephemeral, concurrent, and stateful at once, and persistent and elastic enough to scale with them. TiDB is that backbone,” said Max Liu, Co-founder and CEO of TiDB. “The TiDB Agent State Stack goes a layer further, adding specialized tools for memory, files, and artifacts on top. That is how teams finally move agents from prototype to production without stitching together a separate system for every kind of state.”
The TiDB Agent State Stack combines three core components:
- TiDB Cloud Zero provides a zero-friction database foundation for agent workloads. With a single command, or an embedded skill inside the agent itself, developers provision a fully functional SQL backend to store session histories, task records, tool outputs, retrieval metadata, structured data, and runtime state.
- mem9 offers persistent memory for AI agents, letting them retain information across interactions, recall previous conversations, and build continuity over time. Rather than locking memory into a single model or framework, mem9 runs on TiDB Cloud as a portable memory layer.
- drive9 delivers a persistent workspace for files, artifacts, and agent-generated content. Agents can create, store, retrieve, and reason over documents and outputs while maintaining continuity across sessions and environments.
Together, these capabilities keep memory, operational data, retrieval metadata, files, and execution history connected as agent deployments scale. Most teams currently solve this with three or four disconnected systems, slowing time to market and creating multiple points of failure in production.
This launch strengthens TiDB’s position as a foundational data platform for agentic AI. As model providers introduce their own proprietary memory and workflow features, the TiDB Agent State Stack keeps memory and state portable, governed, and independent of any single AI platform, giving enterprises the control and flexibility to own their agent architecture rather than renting it from a model vendor.
By bringing together TiDB Cloud Zero, mem9, and drive9, the TiDB Agent State Stack gives developers a practical path from prototype to production, so the next generation of enterprise AI agents can operate with the persistence, continuity, and reliability that real-world deployments demand.
About TiDB
TiDB is the AI-native distributed SQL database built for agentic workloads that grow unpredictably. It combines ACID guarantees with native support for transactions, analytics, and vector search in a single engine, giving teams one platform to power AI agents, modern applications, and real-time intelligence without data silos, noisy neighbors, or infrastructure ceilings. Innovators across agentic AI, SaaS, fintech, e-commerce, and gaming, including Manus, Atlassian, Dify, Pinterest, Plaid, and Bolt, rely on TiDB to scale their most demanding workloads. Headquartered in Silicon Valley, TiDB is backed by Sequoia Capital, GGV Capital, Access Technology Ventures, Coatue Management, and others.
Website: https://www.tidb.io/.
View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/tidb-launches-agent-state-stack-at-superai-summit-singapore-gives-ai-agents-production-grade-memory-302797985.html
SOURCE TiDB
