A CRM That Drafts Your Follow-Up Emails

Most CRMs will remind you to follow up. DenchClaw will actually write the email for you — and it'll be good because it already knows everything about the person.

That's the difference. A nudge versus a draft. If you've ever stared at a blank compose window after a sales call, wondering what to say without sounding like a template, you'll understand why that distinction matters.

Here's how DenchClaw's AI email drafting works, and how to make it part of your daily routine.

How AI-Drafted Follow-Ups Actually Work#

When you ask DenchClaw to draft a follow-up email, the agent doesn't start from scratch. It reads:

• Contact history — every previous interaction logged against this person
• Entry document — freeform notes you've written about the contact
• Last interaction date and type — met at a conference, replied to cold email, had a demo call
• Company and role context — what they do, what their company does
• Any custom fields you've set — deal stage, priority, tags

The agent synthesizes all of that and writes an email that sounds like you wrote it after reviewing your notes — because effectively, that's what happened.

Compare that to a template: "Hi , Just following up on our conversation about ..." Templates are obvious. This isn't.

The Action Field Trigger#

The cleanest way to kick off a draft is with an Action field. In DenchClaw, you can add a button-style field to any object called something like "Draft follow-up." When you click it (or say "draft follow-up for [contact name]" in Telegram), the agent activates.

You can also trigger it with natural language directly:

"Draft a follow-up email to Sarah Chen. We met at SaaStr last week, talked about their data pipeline problem. She mentioned their team is evaluating tools in Q2."

DenchClaw will read Sarah's contact record and your note, then generate something like:

Subject: Following up — data pipeline conversation from SaaStr

Hi Sarah,

Really enjoyed our conversation at SaaStr last week. The data pipeline challenges you described — particularly around real-time ingestion at scale — are exactly the kind of problems we've built for.

Given your Q2 evaluation timeline, I'd love to set up a 30-minute technical deep-dive with your team before you finalize your shortlist. Would next week work?

Best, [Your name]

Not bad for a first draft. Usually needs minor tweaking, but the bones are there.

Customizing the Drafting Prompt#

The default drafting behavior works well, but you can tune it. In your DenchClaw config, you can set a drafting system prompt that shapes tone and style:

Draft follow-up emails in a friendly but professional tone. Keep them under 150 words. 
Always reference something specific from the last interaction. 
Never use phrases like "just checking in" or "circling back."
End with a clear single call to action.

You can also override per-draft with inline instructions:

"Draft a follow-up for Marcus at Acme — make it casual, we're already friendly. Reference his comment about the fundraise."

The agent will honor both the system style and your per-request instructions.

AI Drafts vs Templates: An Honest Comparison#

Templates win on consistency and speed for high-volume, low-context outreach. If you're sending 500 emails from a purchased list, use templates.

But for relationship-driven follow-ups — post-meeting, post-demo, post-intro, post-conference — templates consistently underperform personalized emails. The research on this is pretty consistent: response rates drop significantly when people sense they got a form letter.

AI drafts give you the speed advantage of templates with something much closer to the quality of a handwritten email. The key is feeding the agent good context. If your contact records are thin, the drafts will be thin. If your notes are detailed, the drafts will be specific and useful.

This is actually a side benefit of using DenchClaw for AI drafting: it incentivizes you to keep good notes, because better notes = better drafts.

The Morning Routine: Batch Reviewing AI Drafts#

Here's a workflow that works well once you have a few dozen active contacts:

Each morning:

1. Ask DenchClaw: "Who should I follow up with today?"
2. Review the suggested list (based on last contact date, deal stage, custom priority fields)
3. Say: "Draft follow-ups for all of these"
4. DenchClaw generates a draft for each contact in sequence
5. You review, tweak, and approve each one — takes about 2-3 minutes per email
6. Send via Gmail integration

The drafts land in a review queue. You're not approving form letters — you're editing first drafts, which is fundamentally faster than writing from scratch.

On a typical morning this takes 20-30 minutes to process 8-10 follow-ups that would otherwise take 2 hours. The 80% is handled; you add the 20% that only you can add.

Sending via Gmail Integration#

DenchClaw connects to Gmail through the gog skill, which handles OAuth and sends on your behalf. Once authorized, you can:

• Send approved drafts directly from DenchClaw
• Have DenchClaw log sent emails back to the contact record automatically
• Pull in received emails to update contact history

The flow is: draft → review → approve → send → log. All in one place, no tab-switching.

If you prefer to review in Gmail's compose window before sending, DenchClaw can push drafts to your Drafts folder instead. Some people prefer that extra step as a safety net — especially early on when you're calibrating the drafting style.

Batch Drafting for Multiple Contacts#

For post-event follow-up (conference, trade show, batch of intros), batch drafting is where this gets genuinely useful.

After SaaStr, for example:

"I met 12 people at SaaStr. I've added them all with notes. Draft follow-up emails for each of them, referencing our specific conversation notes."

DenchClaw will queue up 12 drafts. Each one will be different because each contact's notes are different. Review and send over coffee.

Without this, those 12 follow-ups either don't happen, happen late (past the 48-hour window where they're still warm), or happen as generic "great meeting you" emails that get ignored.

FAQ#

Does DenchClaw send emails automatically, or do I always review first?

You always review first by default. DenchClaw drafts; you approve and send. You can set up automations for specific low-stakes scenarios (like auto-sending a "thanks for connecting" type note), but anything substantive goes through human review.

What if I don't have detailed notes on a contact?

The draft will be more generic, leaning on company/role context. It'll still be better than a blank compose window, but invest in taking notes after meetings — your future self will thank you when the drafts are actually specific.

Can it match my writing style?

Yes, with a well-crafted system prompt. Feed it a few examples of your actual emails and tell it to match the tone, sentence length, and style. The more examples, the better the calibration.

Does it work for non-English contacts?

Yes. Specify the language in your request: "Draft a follow-up in French for Pierre Dupont." The agent handles it.

What email providers are supported?

Gmail is the primary integration via the gog skill. Other providers can be used by pushing drafts to a file or clipboard and sending manually.

Ready to try DenchClaw? Install in one command: npx denchclaw. Full setup guide →