Now live — agentdb.zizka.ai

Why did your agent
do that?

AgentDB is a database built for AI agents. It stores every decision, links causes to effects, and lets you search or replay anything your agent did.

Get started free →Self-host in 60 seconds

Free tier · No credit card · Self-hostable

quickstart.py
from agentdb import AgentDB

db = AgentDB("agdb_live_xxxx")

# Log what your agent does
msg = await db.log(agent="support-bot",
    event="user_message",
    data={"text": "why is my bill $200?"})

tool = await db.log(agent="support-bot",
    event="tool_call",
    data={"tool": "get_billing", "user": 123},
    parent_id=msg.event_id)   # ← causal link

await db.log(agent="support-bot",
    event="agent_response",
    data={"text": "Found an anomaly in your account"},
    parent_id=tool.event_id)

# Now ask WHY the agent responded that way
chain = await db.why(tool.event_id)
chain.print()
OUTPUT
user_message: "why is my bill $200?"       [14:32:01]
  └── tool_call: get_billing(user=123)      [14:32:02]
      └── agent_response: "Found anomaly"   [14:32:03]

3 events · 1 causal chain · 0.3ms

What Pinecone does not do

Pinecone stores vectors. AgentDB stores what your agent did, why it did it, and what happened next.

🔍
Causal Lineage
See the full chain of events behind any agent decision. Trace back from any output to the exact input that triggered it.
Time Travel
Reconstruct your agent's exact state at any past moment. Useful for debugging what went wrong at a specific time.
🧠
Semantic Search
Search agent history with plain text. Query across thousands of events without knowing the exact keywords.
🚨
Drift Detection
Get notified when an agent starts behaving differently from how it used to. Catch regressions early.
🤝
Agent Handoff
When one agent hands off to another, pass only the relevant context. No more stuffing the full history into a prompt.
📋
Audit Log
Every event is checksummed and append-only. Export a signed audit trail at any time for compliance reviews.

How it compares

CapabilityPineconeMem0AgentDB
Vector / semantic search
Causal lineage (why)
Time travel & replay
Cross-agent fleet intelligence
Behavioral drift detection
Compliance audit export
Self-hostable
Model agnostic

Pricing

Self-host for free forever. Pay for managed hosting when you need it.

Self-Hosted
Freeforever
  • Full feature set
  • Run on your infra
  • Docker in 60 seconds
  • Community support
View docs →
MOST POPULAR
Pro
$49per month
  • 100M events stored
  • 90-day retention
  • 3 projects
  • Email support
Get started →
Team
$149per month
  • 1B events stored
  • 1-year retention
  • 10 seats
  • Priority support
Get started →

Run it on your own server

Full feature set. No account needed. No usage limits. Free forever.

$ docker run -p 8000:8000 agentdb/server
View self-host guide →