A CRM built around one rule.

Message-ID:   <{uuid}@trycustomerluv.com>             — random, stored on the sent row
In-Reply-To:  <{previous step's Message-ID}>
References:   <{step 1}> <{step 2}> … <{previous}>   — the full chain
Subject:      Re: {original subject, verbatim}
Body:         new text, then the quoted prior messages, in text/plain and text/html

Storing the outbound Message-ID on every sent row is what lets a reply from an address you never targeted still find its prospect. The forwarded-to-a-colleague case resolves through References, not the sender. Message-IDs are random, never time-derived, so your sequence doesn't carry an obvious shared prefix.

The one first email the system will ever start on its own initiative: a note to a person you were referred to ("talk to Raj"). It drafts it, you approve it.

The screens below show our own account. Halcyon Skin and the people in it are made up; the products and competitors are real.

Inbox

Every reply, every domain, one place

Replies to every sending domain arrive in one threaded inbox. Each thread carries the product, the competitor detected, the store, and a classification badge. Reply inline and the message goes out from the address the chain started on, with the right headers.

• Mine · theirs · All — opens on your threads, one tap to a teammate's
• Internal notes, visible to the whole account and never sent
• Hand off a thread honestly: a short reply from the new owner, in the same chain, with you on cc

What a reply can mean, and what happens

Sequences

Four short steps, previewed against a real row

One template per product and competitor detected. Variables for the first name, store, competitor, the specific gap, the offer, and your hook. Every rendered step is previewed against a real prospect before the sequence goes live, and per-step reply rate tells you which line is doing the work.

Default cadence is days 0, 4, 10, 14; you can change the gaps, not the chain. Send windows follow the prospect's local business hours, weekdays, with jitter. A physical postal address sits in every signature and a plain-text opt-out line in every step — the CAN-SPAM requirements most cold senders skip.

Domains

Domain health that acts before you notice

One row per sending domain, one colour: healthy, warming, degraded, or paused. The ramp is per mailbox and enforced — 10 a day in week one, then 20, then 30, and nothing raises it past that. A warmup partner runs alongside for the first month; we recommend one and note which you chose. When a domain slides, sends move to its healthier siblings and you get told. Pause stops the domain and tells everyone on the account.

The honest ceiling: about 30 cold sends a day per sender per domain at full ramp. Two people on three domains is roughly 180 new conversations a day. If you need 500+ a day from one seat, this is not your tool — and we'd rather say so here.

Google Postmaster Tools only reports once a domain sends real volume to Gmail, so those two rows fill in later. At launch volumes the decision is carried by the four signals you fully control: DMARC aggregate reports, blocklists, bounce rate and reply rate.

A seat is a person with their own sender address on each domain. Every conversation has an owner — the sender the first email went out from — and the owner replies. Assignment is a rule: round-robin, or by product. If a teammate wants to take over a thread, they do not reply as you. They hit hand off: a short reply from their address, in the same chain, with you on cc. Same References, same conversation, real names. The audit log records who did what. Plans start at two seats and scale from there.