Outbound that can be audited.
Most tools hide how they pick people and what they actually send. We surface every decision, every source, and every guardrail.
Articles of record
These are not aspirations. They are the rules the system enforces before any human sees a draft.
No source, no send
The validateTouch guardrail requires every factual sentence to cite at least one existing SourcedClaim. Dangling citations or unsourced factual content mark the touch blocked - it cannot be approved or sent. The same gate applies to signal drafts.
Buyer Persona, not lookalikes
The Reader builds segments from your actual buyers - titles, pains, triggers, disqualifiers, language patterns - rather than companies that resemble you. Campaigns are created per approved segment.
Copilot is the default
Nothing sends without explicit human approval in an ApprovalBatch. Autopilot is opt-in only, with daily budget caps and a dispatch-time suppression re-check.
Deliverability guardrails
Every address is verified. The circuit breaker pauses on bounce above 3% or complaints approaching 0.1% after a minimum sample. Managed warmed domains by default, with an OAuth lane available.
Global, permanent opt-out
Confirmed unsubscribes write to a global opt-out registry, across all workspaces. Consumer rights - delete, know, opt out - use double opt-in confirmation and are processed through a logged, privileged scrub routine.
Verifiable outcome billing
The $25 fee triggers only on a qualified meeting: accepted RSVP, scheduled start passed, not cancelled or disputed. All accruals are append-only ledger rows with idempotency keys. The outcome fee is capped at 20% of subscription subtotal, and disputes issue compensating reversals.
Compliance, written into the machine
Full operational detail - tenancy isolation, seams, scrub routines - and the DPA live inside the product. The commitments here are intentionally high-level.
Recipient rights
Any recipient can exercise their rights at any time. Full privacy and opt-out tools are available without signing in.
Opt-out confirmations and data requests use double opt-in, so nobody can suppress or delete someone else’s address.
“If a sentence cannot be sourced, it does not leave the system.”
Want to inspect the guardrail? The public demo runs a labeled, illustrative email through the production sourced-claim validator.
Is cold email legal, and what do we enforce?
The questions buyers and their lawyers ask, answered with the mechanisms that run in the product.
What is the no-source-no-send rule?
Rainmaker will not include a factual claim in an outreach email unless the claim can be traced to a verifiable source. If the agent cannot find a source for a claim, the claim is removed from the draft rather than sent unverified. This rule runs on every draft before it reaches the user's approval queue.
How does the human approval gate work?
After Rainmaker drafts outreach, every email appears in a copilot approval queue. The user reviews each draft, edits or rejects it if needed, and must approve it before the email sends. Nothing leaves the account on autopilot.
Does Rainmaker comply with GDPR for cold B2B email?
Cold B2B email is legal under GDPR with a legitimate interest basis when it is relevant to the recipient's professional role and includes a clear opt-out mechanism. Rainmaker includes an opt-out link in every email it sends and suppresses opted-out addresses permanently across all accounts on the platform.
What happens when someone opts out of Rainmaker-powered outreach?
The recipient's email address is added to a permanent suppression list shared across every Rainmaker account. Rainmaker will not contact that address again from any account on the platform, not just the sender who triggered the opt-out.
Is cold email legal in the EU?
Yes, when done correctly. Under GDPR, cold B2B email to a business address is legal with a legitimate interest basis, when the message is relevant to the recipient's professional role, and when the email includes a working opt-out. Rainmaker enforces these requirements automatically.
How does Rainmaker handle bounces and spam complaints?
Every recipient address is verified before a send, and bounce and complaint rates are watched on the sending identity. High bounce or complaint rates trigger an automatic pause of the campaign until a human looks at it.
What happens if the AI writes something factually wrong?
It cannot send. Every factual claim in a Rainmaker email must link to a verifiable source before the draft reaches your approval queue. If no source is found for a claim, the sentence is blocked, not flagged for your review. You can still edit approved drafts before sending, but unverified claims are never presented to you as ready to send.