Every cold call opener you've read about assumes a person picks up. On a growing share of B2B dials, that isn't what happens.

At throxy (YC P25) we've made 3.6 million cold calls. I went back through 355 weekly client reports and 2,764 persona breakdowns to see what those dials turn into. The most common outcome we log is an AI receptionist. It shows up more than twice as often as a human saying they're not interested.

If you run a sales team, your openers were built for a listener who often isn't there. Here's what our data shows and what we're changing.

The best cold call openers all target the same moment

The well-known frameworks are good. Josh Braun's permission-based opener lowers the prospect's guard by giving them an easy way out. Belal Batrawy's mic drop method leads with a sharp problem statement. Gong ran 300 million calls through its system to rank openers by success rate [EXTERNAL LINK: Gong research on cold call openers].

All of them are built around the same half second, when a person hears a stranger's voice and decides whether to keep listening. Every word is designed to win that decision.

That design assumes a human is on the other end.

How often AI receptionists answer cold calls

In 44% of our weekly reports, an AI receptionist is the number one call outcome across the entire account. That's every campaign and every persona we dialled for that client that week, combined.

Three findings stand out:

  • An AI receptionist is the most common outcome we log.

  • It appears more than twice as often as "not interested".

  • Senior marketing titles are the worst for it. Calls to marketing leaders now route through AI almost by default.

Why human-first openers fail on an AI gatekeeper

A permission opener like "if this is a bad time, tell me to bugger off" works because it disarms a person. It hands them control, and that makes them more willing to stay on the line.

A voice agent has nothing to disarm. Its only job is to route you or refuse you. Humour, humility and confidence mean nothing to it. You can't poke the bear when there's no bear.

None of the popular cold calling frameworks cover this. The opener your team spent weeks refining may never reach the person it was written for.

How we're rewriting cold call openers for AI routing

Our SDRs now say the client's company name clearly in the first three seconds of every call.

That line is for the routing model. An AI receptionist listens for a keyword it recognises so it knows where to send you. Give it that keyword early and cleanly, and the call has somewhere to go.

In practice:

  1. Say the company name within the first three seconds.

  2. Say it clearly, so a machine can parse it.

  3. Save the permission line or pattern interrupt for when a person answers.

Writing an opener a machine has to parse is a completely different job to writing one a person judges. Almost nobody in outbound is doing it deliberately yet.

The early results in our data are fascinating. We'll share the numbers in a follow-up once we have enough dials to trust them.

What this means for your cold calling in the AI era

Cold call openers still matter, but your team now needs to get past two listeners: the machine and the person behind it. Before you rework your script, find out what share of your connects are AI receptionists. If your call data doesn't separate them, start there.

throxy runs outbound calling for B2B sales teams and tracks outcomes like this across millions of dials [LINK: How throxy runs cold calling]. If you want to know how your market is answering the phone, book a call with us.

FAQ

What is the best cold call opener in 2026?

It depends on who answers. When a person picks up, permission-based and problem-led openers still work. When an AI receptionist answers, lead with a recognisable company name in the first three seconds so the system can route you.

How common are AI receptionists on cold calls?

In throxy's data from 3.6 million cold calls, an AI receptionist is the most common logged outcome. It appears more than twice as often as a human saying "not interested", and it's the top outcome in 44% of weekly client reports.

Which job titles use AI receptionists most?

Senior marketing titles. In throxy's data, calls to marketing leaders route through AI receptionists more than any other persona, almost by default.

How do you get past an AI receptionist on a cold call?

Give the routing model a keyword it recognises. Say the company name clearly in the first three seconds, then use your usual opener once a human is on the line.