👋 Howdy, GTM Engineer!

I attended one of the best AI in GTM events in NYC last night led by Chris Balestras.

GTM leaders from Clay, Cursor, Kleiner Perkins, and Sensei shared how they evaluate startups to determine:

Check out the strategies shared in my latest LinkedIn post and check out Chris’ newsletter: GTMBA

This week I’m sharing a workflow that gathers data in a way that Clay cannot…

A voice agent that makes calls, has a conversation, and gathers intelligence to programmatically bring back across hundreds or thousands of SMBs.

This is useful for gathering pricing details that Medspas or other similar local businesses may not share on their website, but that you can only get from calling.

Happy Automating!

— Alex Lindahl, GTM Engineer at Clay

GEAR - ‘GTM Engineers Accelerating Revenue’ is Claymation’s series that highlights GTM Engineers and creative workflows & automations they’re building for modern GTM. Reply to this email if you’re interested and tell me a bit about what you’ve cooked up.

This week: Ashish Datta, Partner at SetFive Consulting

We live in a digital-first world, yet a massive amount of high-value competitive intelligence—specifically pricing in industries like aesthetics and healthcare—remains locked behind a phone line.

Medspas frequently withhold pricing from their websites for several reasons:

Traditionally, gathering this data requires a BDR to sit on a phone for hours, navigating IVR trees and manually transcribing numbers. This is unscalable.

The Solution: An automated voice agent workflow that sources targets via Clay, calls them using OpenAI Realtime API, navigates phone trees, and extracts structured pricing data back into Clay or a database.

Before you can dial, you need verified targets. We needed to find Medspas in specific high-volume metros—Boston, Dallas, and Miami—to test the system.

Instead of buying stale lists or scraping manually, we used Clay’s Find Local Businesses source (powered by Google Maps).

The biggest hurdle in AI voice agents has historically been latency. The old stack looked like this: Speech-to-Text -> LLM Processing -> Text-to-Speech. This resulted in awkward pauses that immediately signaled to the recipient that they were talking to a bot.

We leveraged the OpenAI Realtime API to solve this. Because it handles native audio-to-audio processing, we achieved:

To get the AI on the actual telephone network, we utilized Twilio Media Streams. This allows us to pipe the raw audio stream directly from the phone call into OpenAI.

The architecture ensures that the Medspa receptionist hears a responsive potential customer, capable of handling unexpected conversational pivots, rather than a rigid script.

A major friction point in calling local businesses is the Interactive Voice Response (IVR) system (e.g.,"Press 1 for appointments"). These systems listen for Dual-Tone Multi-Frequency (DTMF) signals, not voice commands.

We built a layer on top of Twilio’s callback architecture that allows our AI Agent to:

Getting a human on the line is only half the battle. The goal is data extraction.

Using system prompts designed for structured output, the AI engages in natural conversation to uncover pricing (e.g.,"How much is a first-time Botox unit?"). Even if the pricing is mentioned casually or amidst small talk, the system captures it.

The Output: The agent takes the free-flowing conversation and returns clean, structured data (JSON) containing pricing, availability, and specific conditions (first-time customer deals, minimums, etc.) that can be piped back into your CRM or database.

Small businesses are busy. Often, they don’t pick up. To ensure high success rates without harassing the business with repeated calls, we implemented a full communication cycle loop:

💡 You can also send this data back to Clay via a Webhook so that you can perform further analysis, enrichment, or data transformation on the data collected.

By combining Clay's ability to instantly map local markets with OpenAI's conversational speed, we transformed a manual, high-friction research task into a scalable GTM engine.

Want to hear it in action?

Listen to the Demo at voice2data.setfive.com

If you're interested in implementing voice-based data gathering for your GTM motion, reach out to Ashish: [email protected]

Chainlit is a framework for building your own AI applications that look and feel like ChatGPT. It’s being used by teams at Google, NVIDIA, Microsoft and smaller companies like BuzzFeed & Backmarket.

A customer showed me their app in action and it was pretty slick! It’s like they made their own ChatGPT-style app for their sales team. It’s growing in popularity and has over 50,000 monthly developers building on it.

Primary, a leading seed stage VC firm in New York, is hiring 2 new Associates for their GTM team. Primary’s GTM team works side by side with early stage founders to help them build GTM systems and workflows, run growth experiments, and drive pipeline and revenue.   This role is great for an early career SDR, Sales, or GTM professional that would love the exposure to a VC firm and to hone their growth skills. In this role you will learn pipeline building tools and techniques as well as technical skills.  You must be based in NYC as we are 3-4 days per week in office.  Contact [email protected] to apply.

But also:

I’m putting together the first State of GTM Engineering report with my friends Garrett Wolfe Author of Garrett’s Growth Substack and Maja Voje, Author or The GTM Strategist.

“When I first opened Attio, I instantly got the feeling this was the next generation of CRM.”
— Margaret Shen, Head of GTM at Modal

Attio is the AI-native CRM for modern teams. With automatic enrichment, call intelligence, AI agents, flexible workflows and more, Attio works for any business and only takes minutes to set up.

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉Start for free today