10 Things to Know Before You Deploy Your First AI SDR
AI Summary
SaaStr shares 10 lessons from 10+ months of running AI SDR agents across four vendors, emphasizing that human playbooks must work before automating them. The newsletter also covers how an AI chatbot was built in 25 seconds for $0.20 using Replit, and argues that the SaaS growth slowdown is driven more by budget shifts to Anthropic and OpenAI than by vibe coding replacing enterprise software.
Key Facts
- SaaStr ran AI SDRs for 10+ months across four vendors and found that 90%+ of use cases only need one vendor, and the human playbook must already work before automating it.
- Replit was used to build and deploy a fully functional AI chatbot inside an internal marketing app in 25 seconds for $0.20 in agent credits.
- The SaaS growth slowdown (which began in 2022) is attributed primarily to enterprise budgets shifting to Anthropic and OpenAI, not to vibe coding replacing software.
Contrarian Angle
AI SDRs as Cloning Machines, Not Testing Tools
SaaStr argues AI SDRs should be used to clone already-proven human sales playbooks, not to test new untested copy or cold strategies — flipping the common vendor pitch of using AI for cold experimentation.
Most vendors pitch AI SDRs as a way to test new messaging at scale; SaaStr says this leads to disappointment and that AI should only scale what already converts.
Building Internal SaaS Tools Instead of Buying Them
SaaStr built their own AI VP of Marketing app (10K) on Replit and added a fully functional AI chatbot for $0.20 in 25 seconds, arguing it outperforms 98% of purchasable chatbot vendors.
Contradicts the conventional wisdom of buying best-in-class SaaS tools; advocates for vibe-coding internal tooling as a superior and cheaper alternative.