How to Retrieve Scoped AI Memory Using Recallio's Recall API

Recallio Team

— 2 min read
How to Retrieve Scoped AI Memory Using Recallio's Recall API
How to Retrieve Scoped AI Memory Using Recallio's Recall API
Smart memory isn’t just about storing context, it’s about retrieving exactly what matters, when it matters.

Whether you're building an AI agent, a support bot, or a memory-aware assistant, Recallio's POST /recall endpoint gives you scoped, summarized memory recall in seconds. No vector DB. No fragile RAG. Just memory that works.

In this guide, we’ll show you how to:

  • Query memory by user, project, or team
  • Get summarized recall for low-token injection
  • Filter by tags and TTL
  • Retrieve memory across agents or tools
  • Preview raw vs. summarized output

Basic Memory Recall

Recall the most relevant past entries for a user:

res = requests.post("https://api.recallio.ai/recall", json={
    "user_id": "alex42",
    "query": "What’s their preferred contact method?",
    "summary": True
}, headers={"Authorization": f"Bearer YOUR_API_KEY"})

print(res.json()["summary"])

This will return a compressed summary (if available), scoped to alex42's memory—perfect for injecting into an LLM prompt.

Filtered Recall by Tags or Team

Recall memory related to billing issues, scoped to a specific team:

res = requests.post("https://api.recallio.ai/recall", json={
    "team_id": "support_team_1",
    "query": "What did the user last say about their payment method?",
    "filter_tags": ["billing"],
    "summary": True
}, headers=headers)

You’ll get token-optimized, tag-filtered summaries—ideal for product support, compliance, and multi-user workflows.

TTL-Aware Memory with Context Decay

Recallio automatically decays older, less relevant memory unless marked high-importance.

res = requests.post("https://api.recallio.ai/recall", json={
    "user_id": "u789",
    "query": "What preferences have they shared recently?",
    "decay": True,
    "summary": True
}, headers=headers)

This prioritizes fresh, relevant context over stale logs—perfect for time-sensitive use cases like sales, ops, or legal.

Use Graph-Based Recall (Advanced)

Recall relationships, roles, or entity knowledge with our graph memory search (premium):

res = requests.post("https://api.recallio.ai/graph/search", json={
    "user_id": "alex42",
    "query": "Who leads the Apollo project?"
}, headers=headers)

print(res.json()["results"])

Graph recall connects nodes and relationships (e.g., roles, topics, people) for deeper insights—ideal for research, knowledge assistants, or enterprise use cases.

Compare: Summarized vs. Raw Recall

To fetch raw memory entries:

res = requests.post("https://api.recallio.ai/recall", json={
    "user_id": "alex42",
    "query": "What was discussed about pricing?",
    "summary": False
}, headers=headers)

for m in res.json()["memories"]:
    print(m["content"], m["tags"], m["timestamp"])

You’ll get back original entries, including tags and timestamps - useful for debugging or audit trails.

Next Steps

Read more