Channels/Cold email04

Outreach that writes itself — and tells you which replies matter.

Cold email is mostly an orchestration problem: a data tool, a sending tool, a document full of copy, and a human reading every reply to find the four that count. We generate the sequence from your positioning, ship A/B variants from the first send, and classify every reply by intent so your inbox surfaces the meetings.

Get Early Access See how it works

Step sequence generated by default, at day 0, 3, 7 and 12

Copy variants per step — two subjects, two bodies — live from send one

100%

Of replies classified by intent, with a one-line summary

The problem

You are not running campaigns. You are running a stack.

Apollo for the data, Smartlead for the sending, a document for the copy, a spreadsheet for the variants, and someone reading every reply to find the handful worth a calendar invite. Most of the work is plumbing between tools that know nothing about your product, and the copy suffers because the person writing it is doing five other jobs.

Sequence copy that is a template with a merge field where personalisation should be

A/B testing that never happens because setting it up is a separate project

Every reply read by a human, including the auto-responders and the unsubscribes

No ceiling on generation spend until the invoice arrives

Why this is different

The copy knows what you sell. The inbox knows what matters.

Same sending infrastructure everyone else uses. The difference is what goes into it and what comes back out.

Sequences written from your positioning

Generation reads your BRAIN — profile, positioning, ICP, and your strongest features — so each step argues from something specific and true about your product. Regenerate any single step without touching the rest of the sequence.

Others: a template library and a first-name merge field.

A/B is built in from the first send

Every step ships with two subject lines and two body variants, emitted as spintax so the sending platform rotates them natively. You are testing from email one rather than promising yourself you will set it up later.

Others: one variant and a manual test you never run.

Replies classified, not just counted

Every reply is classified as positive, neutral, objection, or unsubscribe, with a one-sentence summary of what the person actually said. You open the four that want a meeting instead of triaging forty out-of-office bounces.

Others: a reply count and an inbox you still have to read.

Spend ceilings you set in advance

A per-company monthly dollar cap and an independent daily generation cap, so nobody hammering the generate button can burn a month of budget in an afternoon. Both are adjustable per account rather than hard-coded.

Others: unmetered usage and a surprise invoice.

How it works

Bring your own keys. We orchestrate on top.

Connect Smartlead and Apollo

Bring your own API key for each, stored encrypted per company. You keep the vendor relationship, the sending reputation, and the data — we orchestrate on top rather than reselling you someone else’s infrastructure at a markup.

Provision mailboxes with warmup on

Buy sending mailboxes from the dashboard and warmup is configured the moment each one is created: conservative daily volume, a realistic reply rate, weekdays only. We strongly recommend a dedicated sending domain rather than your primary one.

Build the list

Upload a CSV and map your own columns, or run an Apollo people search in the dashboard, save it, and import from it with a hard cap on how many leads you pull. Apollo credit usage is accounted daily so a broad search cannot quietly drain your allowance.

Generate the sequence

Four steps by default, spaced at day 0, 3, 7 and 12, each with two subject variants and two body variants. Written from your BRAIN, not from a template. Regenerate any individual step you do not like.

Edit inline, then launch

Every step is editable in place before anything sends. When you launch, the sequence and the leads are pushed to Smartlead and the campaign goes live under your own mailboxes.

Work the inbox, not the noise

Replies stream back through a signed webhook, get classified by intent, and land in a single inbox view across every campaign. Positive replies are obvious at a glance. Unsubscribes are handled without you reading them.

Pause, resume, or archive

Campaign state is yours to change at any time, and live stats — sends, replies, reply rate — roll up to the dashboard so the channel is comparable against SEO and Reddit.

What you get

Measured in meetings, not opens.

Copy that survives a skim

Sequences that reference your actual wedge, because generation reads a structured profile of your product rather than a company name.

A test running from day one

Two subjects and two bodies per step, rotating natively, so you learn what works while the campaign runs.

An inbox that sorts itself

Positive, neutral, objection, unsubscribe — each with a summary line, so triage takes minutes.

Deliverability handled at provisioning

Warmup switched on as each mailbox is created, with conservative defaults, rather than as a step you remember later.

Predictable generation spend

Monthly dollar and daily call ceilings set per company and enforced before the model runs.

One dashboard with your other channels

Sends, replies, and reply rate sitting next to your SEO and Reddit numbers instead of in a separate tool.

Under the hood

Everything that ships with the channel.

Data & mailboxes

Smartlead bring-your-own-key, stored encrypted

Apollo bring-your-own-key with proxied people search

Daily Apollo credit accounting with a configurable limit

Saved searches with capped import

CSV upload with column mapping

Mailbox purchase with warmup applied automatically

Copy & campaigns

Sequences generated from your BRAIN

Four steps by default at day 0, 3, 7 and 12

Two subject and two body variants per step, as spintax

Per-step regeneration

Inline editing before launch

Pause, resume, and archive

Replies & safety

Signed webhook reply ingestion

Intent classification with a summary line

Unified inbox across campaigns

Automatic retry with a dead-letter queue

Per-company monthly spend ceiling

Independent daily generation-call cap

FAQ

The questions people actually ask.

Do I need my own Smartlead and Apollo accounts?

Yes — this is a bring-your-own-key integration for both. You keep the vendor relationship, the sending reputation, and ownership of your lead data, and you are not paying us a margin on someone else’s infrastructure. Your keys are stored encrypted per company and can be revoked from the dashboard at any time.

How do you protect deliverability?

Warmup is configured on every mailbox the moment it is provisioned, with conservative defaults: low daily warmup volume, a realistic reply rate, and weekdays only. Sending itself runs through Smartlead, so their deliverability tooling applies. We strongly recommend buying a dedicated sending domain rather than sending from your primary one — that single decision protects more than any feature we could ship.

Do I review the copy before anything sends?

Yes. The generated sequence lands in the campaign wizard and every step is editable inline. Nothing sends until you explicitly launch the campaign.

Where do the leads come from?

Either a CSV you upload with your own column mapping, or an Apollo people search you run and save in the dashboard and then import from, with a hard cap on the number of leads pulled. We do not sell you a list.

What stops the AI generation costs running away?

Two independent ceilings, both set per company: a monthly dollar cap on AI generation, and a daily cap on the number of generation calls, so one person repeatedly hitting regenerate cannot exhaust the month in an afternoon. Both are adjustable per account.

How is this different from Smartlead or Clay on their own?

Smartlead is the sending layer and Clay is the enrichment layer, and both are excellent at their jobs — but both expect you to bring the strategy, the copy, and a human to read every reply. We generate the sequence from a structured profile of your product, ship variants automatically, classify what comes back, and put the result on the same dashboard as your other organic channels.

The other channels

SEO

AI-Optimized SEO

Fewer articles, picked from your real Search Console data — written, published, indexed, and refreshed before they decay.

Learn more

Reddit monitoring

Catch the thread where someone asks for what you sell — within ten minutes, in one queue, and reply as yourself.

Learn more

AI Content Studio

Photoreal brand characters, product stills, and short-form UGC scripted from your positioning — yours to download and post.

Learn more

Referrals