SurrealDB Stream #36 - Why your AI agent forgets everything (and what we built to fix it)

Most AI agents can reason. Most can act. Most of them still can't remember - not properly.

On Thursday 23 April, join Tobie Morgan Hitchcock and Matthew Penaroza for the first look at Spectron: SurrealDB's agent memory layer. A structured context layer built directly into the database, not patched together on top of it.

This is a code-first session - expect real queries, honest conversation, and a genuine look at what it actually takes to give your agents persistent, structured memory.

🔎 What we'll cover:

• Why agent memory is broken - the fragmented memory tax, what it costs you in production, and why middleware isn't the answer
• What Spectron is - knowledge graphs, entity extraction, temporal facts, and hybrid retrieval, all inside SurrealDB in a single ACID transaction
• What it means for agentic systems - shared memory across agents, context that persists across sessions, and a layer your stack can actually rely on
• Live code - Tobie shows how it works in practice

🔗 Links:
surrealdb.com/platform/spectron