Draft cold outreach with AI

Coco drafts cold outreach in your voice, grounded in the actual context of the account and contact: recent news, role responsibilities, public commentary, mutual connections. About 4-6 credits per draft depending on research depth; about 32 credits to draft and send a 5-email batch. You review every draft inline, edit the subject, accept or send back. Nothing sends without your approval at the start. Once a sequence pattern is approved and working, you can let Coco run it autonomously inside your guardrails. The result is cold email that reads like a colleague wrote it after research, because that's what happened.

Try Coco for free → · 1,000 credits free · no card · ~2-min setup

The cold-email tradeoff most teams are stuck inside

Cold email is a tradeoff between volume and quality, and most teams are stuck on the wrong side of it.

Personalize every email properly (pull the research, find the hook, write in a way that signals you actually read the person's LinkedIn) and you spend about 15 minutes per email. An SDR with two hours of inbox time a day sends 8 personalized cold emails. The response rate is good; the volume isn't enough to fill pipeline.

Template and spray and you can send 200 a day. The response rate is bad. Half get flagged as suspicious, your sender reputation degrades, and the few that respond are probably the wrong fit anyway.

Most SDR teams oscillate between the two and end up with the worst of both: partial personalization that takes 5 minutes and reads like partial personalization.

Coco's bet is that with the research already done and the draft already written, the bottleneck isn't the work; it's the judgment. You can spend 30 seconds reading a draft, edit a subject line, and approve a send in less time than it takes to write a single cold email from scratch. The quality stays high because the research stays high. The volume goes up because the writing time disappears.

How Coco drafts a cold email

The workflow runs end to end in one chat. You point Coco at the contact (or a list); Coco proposes the plan with the credit cost; you approve; Coco runs research and drafts in parallel.

The drafting sequence per contact:

Pull the research. Coco runs an account research pass: public commentary, recent activity, role responsibilities, mutual connections, problem signals. About 5-8 credits per contact at standard depth.
Identify the hook. Coco looks for the most specific, most recent, most relevant anchor: a podcast quote, a LinkedIn post about the problem you solve, a hiring signal that implies the work is on their roadmap.
Draft in your voice. On the Founder tier and above, Coco uses your voice training (sample sends, tone preferences, structure rules). Without explicit training, Coco infers from your sent history in Gmail. The draft includes a subject line, opening hook, value proposition, and a clear single-step ask.
Surface for review. The draft appears in chat with the hook source linked back. You edit inline, request a rewrite, or approve. About 4-6 credits per drafted email all-in.
Send on approval. Approved drafts queue in Gmail drafts under your sender. Send goes through Coco only after your explicit approval the first time you run the pattern.

Try Coco for free → · 1,000 credits free · no card · ~2-min setup

What "grounded in context" actually means

The voice-anchor example from the landing page shows what this looks like in practice:

Subject: the CRM comment on Lenny's

Sarah — heard you mention your CRM was "held together with duct tape." We've been working on...

That subject line isn't generated from a template library. It's pulled from a specific podcast quote on a specific episode that Coco's research surfaced. The opening anchors to the same source. Sarah opens the email and recognizes the reference. The credibility cost of "this is automated" is zero, because the email reads like someone who actually heard the podcast.

The signals Coco draws from:

Recent funding announcements (Crunchbase, news mentions, LinkedIn announcements).
Hiring signals (LinkedIn jobs board, careers page changes, public announcements).
Public commentary (podcasts, LinkedIn posts, Twitter/X, conference talks, blog posts).
Role responsibilities (LinkedIn title and bio, company team page).
Mutual connections (LinkedIn integration surfaces warm paths).
Firmographic context (Apollo or ZoomInfo for company size, stage, industry).

If Coco can't find a real hook (no recent activity, no public commentary, no specific signal) the draft says so. Better to surface "no strong hook available, here's a generic opener you should rewrite" than to invent a quote. Coco won't fabricate context to fill a draft.

Approval, editing, and the credit math

Every draft is reviewable. Every send gates on your approval the first time. You can:

Edit any field inline (subject, body, CTA).
Send back with a note ("the tone is too formal," "open with the funding angle instead").
Approve a batch in one click after individual review.
Skip a contact with a reason Coco remembers for next time.

The credit math is transparent on the plan card before anything runs. A 5-email batch is roughly 32 credits all-in: about 5 credits per email for research and drafting, plus 7 credits for orchestration (queue, send-through, log, audit). You approve or change the plan.

From draft to sequence

A single cold email can extend into a multi-touch sequence. Coco drafts a first-touch plus two follow-ups in one pass, anchored against the same research. Each touch has a distinct angle: the first is the cold hook, the second references the first and adds a value proposition, the third proposes a specific next step.

Every touch is drafted before any ships. You approve the sequence as a whole or touch by touch. For ongoing reply management, see how Coco automates follow-ups.

Why this beats the AI SDR model

The AI SDR category (11x's Alice, Artisan's Ava, and the rest) is built on a different premise. They're optimized for autonomous outbound at volume. The pitch is that you replace SDR headcount with an AI that researches, drafts, and sends without you in the loop.

That premise works for some teams. If outbound volume is your only bottleneck and your brand can absorb the deliverability tradeoff of high-volume autonomous sending, an AI SDR is a credible choice. Coco won't try to outrun it on volume.

But for most teams shopping AI cold-email tools, the bottleneck isn't volume; it's the personalization tax that makes any meaningful volume unaffordable in human time. Coco's model is "fewer emails, better drafts, your approval at the start, no spray." Quality positioning, not volume positioning. The voice is yours; the research is real; the send only happens when you've read the draft and approved it.

For teams running 11x or Artisan already, Coco doesn't compete; it operates as the layer around them. Deeper research on high-value accounts, CRM hygiene on the contacts the AI SDR touches, follow-through on booked meetings, contextual reply drafting the autonomous bot wouldn't handle. See the 11x and Artisan integration pages.

Frequently asked questions

Can Coco match my exact voice?

Yes. Voice training is available on the Founder tier and above. You provide sample sends, Coco learns the pattern, and drafts read like you wrote them. Without explicit training, Coco infers from your sent-history in Gmail; the inference is decent but explicit training is sharper.

How long does a single draft take?

About 1-2 minutes per draft including research, at standard depth. The plan card shows the time estimate before you approve. Bulk batches run in parallel; a 10-email batch typically takes 4-8 minutes total.

Can I run a multi-touch sequence?

Yes. Coco drafts the full sequence (first touch + 2-3 follow-ups) before any ships. Approval-gated at the sequence level or per-touch, whichever you set up. Replies on the sequence get handled by the follow-up workflow.

What if the personalization is wrong?

Send the draft back with a specific note. Coco re-drafts with the constraint and remembers it for the rest of the batch and for future runs in the same workflow. The drafts get sharper as you give feedback.

Will Coco send if I'm away?

No, unless you've explicitly approved an autonomous send rule for a specific workflow. Default is approve-before-send; graduate proven patterns to autonomous later. The approval gate is the brand, not a configuration toggle.

Get started

Hand Coco one target list this week: five to ten contacts you'd otherwise spend two hours personalizing. Approve the plan, read the drafts, decide whether the research and the voice match what you'd have written yourself.

Try Coco for free → · 1,000 credits free · no card · ~2-min setup

Or book a walkthrough → if you'd rather see the drafting flow run on a real account first.