streaming · live agent narration
Watch the AI agent descend the waterfall, token by token.
An AI agent narrates every step of the cascade — which tier ran, which signal fired, what was verified, and why it exited. No black box.
cascade-agent.live
stream · tier-0-exit · 142ms
enrich linkedin.com/in/jane-doe
PARSEExtracted name Jane Doe, domain acme.com, 8 permutation candidates
T0.1Hit CommonCrawl, GitHub commits, cert transparency, ATS job board · 14 signals collected
T0.2MX fingerprint: standard Google Workspace · no security gateway · no catch-all
T0.3AI judge fused signals → top candidate jane.doe@acme.com · confidence 0.96
T0.4SMTP probe (priority MX) → 250 OK · mailbox exists
LEARNWriting pattern first.last@acme.com to domain memory · next call free
→ result
jane.doe@acme.com · verified · T0 exit · cost $0.00 · 142ms
spec
Inspect the agent's work. Trust it on the data.
Every enrichment streams its decision path back. Trust signals are reproducible. The agent can be re-run on any row to see the same descent.
// uptime
99.9%
Five tiers operational. SOC 2 in progress. SLA available on enterprise.
// free tier
200 rows/mo
No credit card. Use HTTP API, Python SDK, or n8n template.
// median
142 ms
T0 exit latency for cached domains. Cold path runs through tiers as needed.
// price
fraction
Pay for as much waterfall as the row needs. Free tiers run first.
Stream the agent on your first row, free.
200 enrichments/month · no credit card · cancel anytime.