Persistent memory for AI agents.
Built on Model Context Protocol.
Keep the decisions, preferences, and task context your agents need. HeurChain gives your AI tools a shared place to save useful memory and find it again across sessions.
From context to continuity
See the connections. Carry context forward. Give memory a useful lifetime.
Explore how memory connects
Explore a simulated workspace of memories, agents, and connections. Click a node to inspect the example.
When should you use persistent agent memory?
Use HeurChain when an agent needs saved decisions, project conventions, or handoff notes from earlier sessions. Store the context with tags, retrieve it in the next session, and export a readable copy. Your application still decides what to save and which retrieved facts to trust.
Give different memories different lifetimes
Choose a namespace when you save a memory. Its decay policy affects how it ages in the active working set.
Hot operational context
Current task state, debugging notes, and short-lived operational context. Use this tier for details that become less useful quickly.
Reference knowledge
Cross-session knowledge, summaries, and working notes. Activation accounts for age and retrieval activity.
Agent identity
Longer-lived project conventions, persona notes, and preferences. This tier decays more slowly; it does not enforce instructions or guarantee permanent retention.
How memory flows through HeurChain
Your integration stores context, searches memory, and brings relevant results into the next task.
Connect, save, and search.
- →BM25 + Qdrant hybrid search, fused via Reciprocal Rank Fusion
- →Save decisions, preferences, and handoff notes with tags
- →Tenant-scoped access with your workspace credentials
- →Remote MCP over SSE for compatible clients, plus a REST API
{
"mcpServers": {
"heurchain": {
"url": "https://api.heurchain.com/mcp/sse",
"headers": {
"X-HeurChain-API-Key": "YOUR_API_KEY",
"X-HeurChain-Tenant": "YOUR_TENANT_ID"
}
}
}
}curl https://api.heurchain.com/store \
-H "Content-Type: application/json" \
-H "X-HeurChain-API-Key: $HEURCHAIN_API_KEY" \
-H "X-HeurChain-Tenant: $HEURCHAIN_TENANT_ID" \
-d '{
"text": "Use UTC for billing timestamps.",
"agent_id": "my-agent",
"namespace": "notes",
"tags": [
"decision",
"project-context"
]
}'curl https://api.heurchain.com/search \
-H "Content-Type: application/json" \
-H "X-HeurChain-API-Key: $HEURCHAIN_API_KEY" \
-H "X-HeurChain-Tenant: $HEURCHAIN_TENANT_ID" \
-d '{
"text": "billing timestamps",
"tags": [
"project-context"
],
"limit": 5
}'Memory for your agents. Pick your tier.
Hosted memory from $5/month. Plans include unlimited agents within storage, operation, and request limits.
Explore the public implementation and its documented setup requirements.
- Source and examples on GitHub
- Separate from the hosted release
- Check licensing and dependencies
1 tenant · 1 human seat · unlimited agents within plan limits.
- 25,000 vectors
- 50,000 operations / month
- 20 requests / minute
- MCP and REST access
- Markdown vault export
One seat included. Additional seats are $9.99/month each.
- 1 tenant per subscription
- 2,000,000 vectors
- 5,000,000 operations / month
- 600 requests / minute
- Unlimited agents within plan limits
Discuss requirements for a deployment that fits your organization.
- Custom capacity requirements
- Dedicated infrastructure options
- Support and SLA requirements
See pricing for current quotas, trial details, and billing terms.
Download a Markdown vault snapshot from the dashboard. See Terms for cancellation and retention policies.
Agents in the same workspace share its memory. An agent label is not a separate access boundary.