Use Cases › Multi-Agent Systems
Multi-agent systems fail when agents disagree about facts. Verified memory makes shared knowledge trustworthy.
Get API Key →Research agent stores "Q3 revenue was $4.2M." Writing agent uses it. Fact-checking agent finds the real number was $4.7M and stores a correction. But writing agent already published the first number. Two "facts" in memory. No flag that they contradict.
Three agents. Different permissions. Governance enforced.
Write. Block. Verify. Governance is the product.
Agent A stores fact, Agent B stores conflicting fact, nobody notices.
Agent C reads a fact that Agent D superseded 30 seconds ago.
All agents read/write everything, no policy on who writes authoritative facts.
// Research agent writes
await client.ingest({
namespace_id: 'report_q3',
content: 'Q3 revenue was $4.2M based on preliminary data',
source_type: 'research_agent',
});
// Fact-checker finds the real number
await client.ingest({
namespace_id: 'report_q3',
content: 'Q3 revenue was $4.7M based on audited financials',
source_type: 'factcheck_agent',
});
// VerifiedState automatically detects the conflict
const conflicts = await client.query({
namespace_id: 'report_q3',
query: 'Q3 revenue',
include_conflicts: true,
});
// conflicts[0].conflict_type → "contradicting_value"
// Both assertions preserved, conflict surfacedWhen agents share a namespace, every contradiction is surfaced before it propagates. No silent overwrites. No stale reads.
{
"receipt_id": "rct_7f3a...",
"assertion": "User prefers dark mode",
"status": "verified",
"confidence": 0.94,
"signature": "ed25519:a8c3f1...",
"verified_at": "2026-04-06T14:32:01Z",
"chain_position": 42
}Free: 50,000 assertions/month. No credit card. No trial expiration.