SmartDrip

Prospecting is repetitive, low-leverage work: finding leads, enriching them, writing outreach, and chasing replies. Doing it across more than one organization adds a harder requirement, which is that each org's data has to stay completely separate.

I wanted a CRM that automated the busywork with AI but was multi-tenant and safe by construction, not as an afterthought.

Every row is org-scoped through row-level security, across owner, admin, member, and viewer roles. Leads enter by CSV, CRM sync, or prospect search, pass a compliance gate, and run through an async enrichment queue that resolves verified contact details. An AI step drafts four channels of outreach; those drafts become a drip campaign a human activates; a scheduler sends from the user's own mailbox under send-windows and daily caps. Replies pause the sequence and get an AI-drafted response that a human approves before it goes out.

Org-scoped row-level security is the foundation, not a feature bolted on later, because multi-tenant safety has to be structural. I send from each user's own mailbox rather than a bulk provider, which protects deliverability and trust.

The reply loop drafts with AI but never auto-sends. The system can auto-send, and I left it off on purpose: the reply is the riskiest surface, and a human belongs on it. The async queues are idempotent and self-healing, because prospecting is bursty and external providers fail.

The AI writing the message is the easy part. The real product is the guardrails around it: tenancy, deliverability, compliance, and reply handling.

Keeping a human on the reply gate, even where I could have automated it, was the right call and not a limitation.