Memory
Memory that survives the run.
The graph asks for memory. The host keeps it. remember, recall, and search are calls on the run — not a weekend of schema design.
Your workflow decides
- Which facts should persist beyond the current run
- When earlier context should be recalled
- What knowledge a node should search for
Linea provides
- A managed fact store for each workspace
- A hosted retrieval path for shared knowledge
- The same memory from the builder and SDK
Example
Remember, recall, and search
The same memory from the builder and the SDK. One store for the agent, not one store per graph.
await memory.remember("billing", {
plan: "annual",
last_invoice: "failed",
});
const facts = await memory.recall("billing");
const policy = await memory.search(
"billing",
"retry policy for annual plans",
);How it works
Persistent context, available on the next execution
Store structured facts for future runs and give every graph a consistent way to search shared knowledge from the host.
01.
Remember structured facts
Save durable information such as a customer plan or account state as records the next execution can read.
02.
Recall context on the next run
A later execution can retrieve what an earlier one learned, so the agent has continuity across conversations and triggers.
03.
Search workspace knowledge
Any graph can retrieve policies or product guidance through one hosted search path.
Compare
Add memory without operating another service
Linea keeps facts and workspace knowledge on the host, so every graph behind the agent reads the same store.
| Self-managed memory | Linea Memory | |
|---|---|---|
| Where facts live | A database and schema you maintain | A managed workspace store |
| How a node reads them | Custom integration for each graph | Shared remember, recall, and search APIs |
| Workspace documents | A retrieval pipeline you operate | Hosted workspace search |
| Builder vs SDK | Separate integrations can drift | One memory layer across both surfaces |