Matteo Tittarelli tested replacing human BDRs with AI at scale-up companies like Twamp. A hybrid model of humans using AI tools beat fully automated systems every time, even after attempts to optimize prompts or build better autonomous systems.
In, in a scale-up, absolutely. And proving the point, if you look again, you were asking about Twamp, if you look at Twamp's organization and many other organizations and you look at outbound to go back to this AI SDR, BDR, like software example, they have BDRs, they've got 100 BDRs, right? And we tested and we invested in trying to replace them and very rigorous analysis proving that a hybrid model of humans plus AI outperform a full AI for now in the context of LLM. And believe me, I tried, I tried beating that hybrid setup with a full auto and I failed. And other people have come to me and says, oh gee, like there must be a way to like, there must, we can like improve the prompt and they have failed.
Roll Claude out from one laptop to the whole GTM org | Sam Gong & Zach Vidibor
