How do you set up Chloe's voice agent without wasting your first week?

Close says Chloe takes under an hour to set up. That's true, and it's also the part that gets teams in trouble.

An hour is how long it takes to configure the agent. It's not how long it takes to know what you should be telling her to do. Skip that part, and you'll be back in the settings by Thursday, rewriting a script you should've gotten right the first time — and re-explaining to your team why the meetings on their calendar don't quite make sense. The gap between those two timeframes is where almost every rocky Chloe rollout actually starts.

This is the single most common way a Chloe rollout goes sideways, and it's entirely avoidable. The fix isn't more caution or a longer checklist. It's doing the thinking in the right order, before the setup screen instead of after the first bad batch of calls has already gone out to fifty real leads who now have an odd first impression of your business.

What should you define before you open the setup screen?

Your qualification criteria, in plain language. What makes a lead worth a meeting, versus worth a polite no. Most teams have this in their head and nowhere else, which is fine for a human rep who can improvise around gaps and useless for an AI agent who needs it written down in a way that leaves no room for interpretation.

Write it as a short, ordered list: the two or three things that have to be true for a lead to be worth a meeting, in the order you'd actually check them on a real call. If you can't produce that list in ten minutes, that's a sign your team doesn't actually agree on it yet, and that's worth fixing before Chloe amplifies the disagreement across hundreds of calls instead of just a handful.

Get specific about the edge cases too. What happens with a lead who's a perfect fit on paper but vague about timeline? What about someone who clearly has budget but is shopping three competitors at once? Decide those answers now, in a document everyone on the team can see, instead of discovering three different answers three different reps would have given if you'd asked them separately.

What goes into the actual script?

Write it the way your best rep talks, not the way a script usually reads. Chloe needs the questions, the order, and what to do with each answer. The script is the process — if it's vague, every call she makes will be vague too, and that vagueness compounds every single time she picks up the phone.

Include the disqualifying answers explicitly. If someone says their timeline is 'sometime next year,' does that disqualify them or just move them to a different follow-up cadence? Decide that in the script, not in a Slack message after the fact, three days into the rollout when someone notices it's already happened a dozen times and nobody agreed on what should've happened instead.

Which leads should Chloe touch first?

Don't point her at your whole database on day one. Start with one segment — inbound leads from a specific form, or a batch of old opportunities that never got a real follow-up. Watch what happens closely. Fix the script based on real calls before you scale it to anything bigger than the test group you started with.

A segment of fifty to a hundred leads is usually enough to surface the real problems with a script without exposing your whole pipeline to them. Bigger than that, and you're just generating more data you don't have time to review carefully, which defeats the entire purpose of starting small and deliberate in the first place.

Who should own the first week?

Someone specific, not 'the team.' That person listens to early call transcripts, tracks how many calls end in a clean qualified or not-qualified outcome versus a confused one, and owns adjusting the script daily for the first week. Without a named owner, this step quietly doesn't happen, and nobody notices until the pattern has already repeated fifty times.

This doesn't need to be a full-time responsibility. It needs to be someone's explicit responsibility, with an hour or two set aside daily for the first week specifically to listen and adjust. That small, deliberate investment up front is what separates a smooth rollout from a rocky one that takes three extra weeks to stabilize.

What should the CRM look like before Chloe touches it?

Every field she needs to read from, and every field she's expected to write to, should already exist and mean something consistent across your whole team. A voice agent that's supposed to log 'qualified' into a field that half your team has been using loosely for months will just make that inconsistency visible faster, not fix it on its own.

This is usually a half-day task if your CRM was already reasonably organized, and a much bigger one if it wasn't. Either way, it's worth doing before setup, not after the first week of calls exposes exactly where the gaps were sitting the whole time, waiting to cause confusion at scale.

What actually causes a re-do?

Turning Chloe loose before your stages and fields mean something consistent. If two reps define 'qualified' differently, Chloe will make that inconsistency worse at scale, not better. Fix the definitions first. The hour of setup is genuinely the easy part — it's everything that has to be true before that hour that actually determines whether the rollout goes smoothly.