No Memory, No Harness: Why the Database Is the Last Line of Defense — Kay Malcolm, Oracle
Kay Malcolm from Oracle describes a problem: when her distributed development team uses AI tools, each individual becomes faster, but the team as a whole doesn't become more productive.
The reason is that Git only records which code changed, not why. Malcolm explains that an AI agent needs memory — a central nervous system carrying context between the model and the tools it uses. She distinguishes five memory types: short-term memory within a session, long-term memory across sessions, episodic memory for past events, procedural memory for steps taken, and semantic memory for meaning. The problem becomes acute when data is scattered across multiple database types: each specialized database adds overhead in security and updates. Malcolm demonstrates with four volunteers on stage — one for each database type — who cannot agree on what is true. The solution is to store every memory type in one place so the agent doesn't guess wrong and waste AI tokens.
Vibekollen prepared this summary with AI from the original publication. The content belongs to AI Engineer.
More to read
Lumaril, which develops AI software for optimizing industrial production by connecting to existing control systems, raised a $66M Series A at a $400M valuation
Techmeme 4 h ago
DeepSeek open sources Super-Agent as a research prototype, a proactive coding agent it says can monitor group chats for coding issues and suggest solutions
Techmeme 5 h ago
Insurance software maker Guidewire agrees to acquire Hyperexponential, which makes AI-powered insurance underwriting and pricing software, for $1.7B in cash
Techmeme 5 h ago
One word for AI agents in 2027
Nate Herk 9 h ago