Memory for
AI Agents.
Sub-200ms.
Give your agents persistent memory with 4 API methods. Store, recall, search, forget — in under 200 milliseconds.
Works with your stack: · LangChain · LlamaIndex · CrewAI · AutoGen · OpenAI · Vercel AI SDK
Hey Grid Bot, what's BaseGrid?
BaseGrid is the AI memory API I'm running on! Sub-200ms retrieval, 99.9% uptime...
Meet Grid Bot
Built on BaseGrid
Our AI assistant runs entirely on BaseGrid infrastructure. It's not a demo — it's a real production bot serving thousands of conversations.
"We eat our own dog food."
Grid Bot proves BaseGrid works at scale. If it's good enough for our bot, it's good enough for yours.
60 seconds to persistent memory
No infrastructure to manage. No vector databases to configure. Install, initialize, and your agent remembers.
Install the SDK
One package. Node.js, Python, Go, or React Native. No dependencies to wrangle.
Store memories
5 memory types — conversational, factual, procedural, episodic, semantic. Auto-classified.
Recall in <200ms
Hybrid semantic + keyword search. Your agent gets context instantly.
LATENCY COMPARISON
Simple API.
Powerful Results.
Add memory to your AI agents in minutes. Our SDK handles the complexity of vector search, caching, and retrieval so you can focus on building.
Works with
1import { BaseGrid } from '@basegrid-io/core'
2
3const client = new BaseGrid({
4 apiKey: process.env.BASEGRID_API_KEY
5})
6
7// Add a memory
8await client.memory.add({
9 agentId: 'support-bot',
10 content: 'User prefers email communication',
11 metadata: { userId: '123' }
12})
13
14// Search memories
15const results = await client.memory.search({
16 agentId: 'support-bot',
17 query: 'How should I contact the user?'
18})
19
20console.log(results[0].content)
21// Output: "User prefers email communication"Ship agents that remember.
Get your API key in 60 seconds. No credit card required.
npm install @basegrid-io/core