LAUNCH
Clean a list before launch
Verify every email on Friday’s list, dedupe against HubSpot, remove customers, and return the send-ready CSV with the risky rows separate.
ONE MISSION · EVIDENCE ATTACHED
⟨ LEAD VERIFICATION AGENT ⟩
Nothing reaches the sequence unverified.
It verifies emails and phones in bulk, deduplicates against your CRM, suppresses customers, competitors and do-not-contact records, scores fit against your ICP, and hands back a list your sender domain and your reps can trust, with the risky rows split out where you can see them.
POST /api/v1/agents/verificationINPUT · ANY CONTACT LISTCONSOLE · REST · MCP
⟨ WHAT IT READS. WHAT IT DOES. ⟩
Capabilities with the receipts.
Bulk verification
email_verify · find_phone · list_enrichEmails checked for deliverability and phones for validity across the whole list, with the result and the date stored on the row.
Dedupe and suppression
list_hygiene · crm_upsert_contactsRows matched against existing CRM records on identity keys, then customers, open deals, competitors and do-not-contact lists held out.
Fit scoring
list_score · company_lookup · person_enrichEvery row scored against the ICP you describe, so the sequence starts with the best fit and the long tail waits its turn.
The honest split
run_code · list_export · deliver_csvDeliverable, risky and unknown rows reported separately with counts computed exactly, so the send decision is made on numbers.
⟨ CAPABILITY RADAR ⟩AGENT: VERIFICATION
⟨ MISSIONS IT RUNS ⟩
Described in plain language. Delivered end to end.
HYGIENE
CRM hygiene sweep
Every contact created in the last 90 days: verify, flag duplicates, and propose merges as an approval batch.
ONE MISSION · EVIDENCE ATTACHED
INBOUND
Score the inbound
Score this week’s 300 signups against our ICP, verify their work emails, and route the top 30 to the right owners.
ONE MISSION · EVIDENCE ATTACHED
What comes back: A send-ready list with the counts that justify it: verified, deduplicated, suppressed and scored, with the risky rows named.
⟨ GOVERNED AUTONOMY ⟩
Autonomous where it earns it. Governed everywhere.
A hard cost cap on every run
Each run reserves budget before it executes and settles after - the cap is enforced by the ledger, never advisory.
Evidence behind every finding
Numbers come from tool output only, and every finding cites the capability that surfaced it.
Consequential actions wait
Anything that spends, publishes or reaches out routes through your approval policies before it moves.
Three ways to call it
The console for teams, the CLI in your terminal, REST with an API key, MCP inside your assistant - the same agent on the same meter.
⟨ THE ROSTER ⟩
Other AI Agents
⟨ DEPLOY THE VERIFICATION AGENT ⟩
Give it the objective. Read the evidence.
Every workspace ships with all seven agents and the full data stack behind them. The first mission runs minutes after signup.