Our AI Worker Booked 12 Meetings This Week — Here's the Setup

Last week, our sales worker booked 12 meetings from inbound leads. No human wrote a single follow-up email. No one checked a CRM or copied a template.

The worker handled outreach, answered product questions, dealt with objections, and sent calendar links on its own. We built this on Spinnable, our own platform. This is our actual pipeline, with real leads and real revenue on the line.

I want to show you how we set it up, because the whole thing took less than an hour.

The Problem: Leads Were Dying in Our Inbox

Spinnable has a small team. Three people run the company. None of us had the bandwidth to follow up on every demo request and trial signup within a reasonable window.

I've built companies before (my last one, Unbabel, grew to 400 employees), and I know the cost of slow follow-up. Leads go cold fast.

The pattern repeated every week. A lead would sign up, receive an automated welcome email, then hear nothing for two or three days. By the time someone wrote a personal follow-up, the lead had already evaluated two competitors and picked one.

I measured our average first-response time over a four-week stretch: 52 hours. After setting up the sales worker, that dropped to 8 minutes.

How We Configured the Sales Worker

The whole setup took about 45 minutes. I defined the worker's role, connected the tools it needed, and wrote the instructions that govern how it communicates. No code. No integrations to maintain.

Role definition. I scoped the worker as an SDR for Spinnable: qualify inbound leads, answer product questions, book demo meetings. Nothing else. No pricing negotiations, no support tickets. A focused worker performs better than one trying to cover everything.

Trigger. Every time someone signs up for a trial or requests a demo, our app fires a webhook. The worker picks that up as its cue to start. No polling, no delay.

First action: research the lead. Before sending anything, the worker pulls the lead's LinkedIn profile and company information. It checks their title and company size, then cross-references with their signup details. This takes about 30 seconds and gives the worker enough context to write something specific.

Second action: send a personalized email. Within 10 minutes of signup, the lead gets an email. The prompt I wrote for this step looks something like:

"Write a short email to {first_name} at {company}. Reference their {industry} and the fact that they signed up for {signup_type}. Ask one specific question about their current workflow. Keep it under 100 words. No greeting filler. Sign off as the Spinnable sales team."

The result reads like a message from a real teammate. The worker has context about who it's writing to and strict guardrails about tone and length.

Follow-up sequence. If the lead doesn't reply to the first email, the worker follows a timed sequence: a check-in on day 2, a value-focused nudge on day 5, and a final reach-out on day 8. Each follow-up references the previous message and adds a relevant use case or a concrete example of how a similar company uses the product.

After week 1, I shortened the sequence from five touches to four. The fifth email felt like too much, and reply rates on that last touch were near zero.

In-thread conversations. This part surprised me. Several leads replied with questions before booking: "Does this integrate with HubSpot?" or "Can it handle multi-language emails?"

The worker answered those questions accurately in the same email thread, then steered the conversation back toward booking. I didn't build a separate Q&A flow. The worker's knowledge base and role instructions handled it.

Booking. Once a lead expressed interest in a call, the worker sent a calendar link connected to my schedule. The lead picks a slot and both of us get a confirmation. No back-and-forth about availability.

The Numbers After One Week

Here's what the worker produced in its first full seven days:

• 147 personalized emails sent across initial outreach and follow-ups
• 41 replies received (28% reply rate)
• 12 meetings booked
• 3 leads flagged as "not a fit" by the worker itself

For comparison, our manual process over the previous month averaged 2 meetings per week. I was doing that outreach myself, pulling 4 to 5 hours away from product work every week to write follow-ups and chase replies.

The results weren't flawless. 106 emails got no reply at all. Some of those leads were never going to convert regardless of who emailed them. A few of the worker's messages could have been sharper.

One lead told me the follow-up felt "a bit generic," which was fair feedback. I adjusted the prompt afterward to pull in more company-specific details for the second and third touches.

The three "not a fit" flags were all accurate. Two leads were university students researching AI for coursework. One was a competitor's employee signing up to poke around. The worker caught these from LinkedIn data and email domains, then skipped the follow-up sequence entirely. That saved me from three demo calls that would have gone nowhere.

Running for Four Weeks, One Adjustment

After shortening the email sequence in week 1, I haven't changed the configuration. The worker has been running the same playbook for four straight weeks.

Monthly totals: 43 meetings booked, with a consistent 26-30% reply rate on initial emails. The worker picks up weekend signups and holiday signups with the same speed as a Tuesday afternoon. I open my calendar on Monday mornings and find meetings already scheduled.

A three-person team doesn't need a full-time SDR hire. We needed something that could respond fast and stay consistent with the repetitive outreach None of us had capacity for.

The 45 minutes I spent configuring this replaced roughly 20 hours of manual work per month. If you're running a small team and your inbound leads sit untouched for days, you can build something like this in under an hour on Spinnable. The setup is the easy part. The hard part is admitting you need it.