At SaaStr AI London, Amelia and I delved into our AI SDR (Sales Development Representative) journey, sharing all our emails, data, and performance metrics. The reaction was overwhelming, yet the top objection was, “This won’t work for me—I lack your scale, data, and 10 years of history. ” That’s false. If you have customers, revenue, and any size database, AI agents can work for you. You don’t need vast data or history, but rather a solid methodology. After sending 60, 000+ hyper-personalized emails, booking 130+ meetings automatically, and generating 15% of our London event revenue via AI agents (potentially 50% by SaaStr AI Annual 2026), here are our five biggest learnings: 1. **AI Agents Do What Humans Won’t:** Human SDRs avoided following up on return attendees or “ghosted” leads pursuing smaller deals, focusing instead on high-value sponsors. Despite incentives and monitoring, they didn’t deliver. AI agents tackled these neglected leads, generating 15% of London ticket revenue and achieving a 70% open rate on ghosted leads. AI SDRs succeed by doing tedious, low-priority work humans skip. 2. **Hyper-Personalization at Scale Works—“Pretty Good” Is Enough:** Humans sent 75-300 personalized emails monthly; AI sent nearly 60, 000 in six months—32x the output. AI emails are modestly personalized (level 3 to 6 out of 10), referencing company names or recent activity but not crafting elaborate messages. Consistent, decent personalization at scale beats sporadic human brilliance. 3. **Train Agents Like New Hires:** AI SDR products don’t automatically generate revenue. You must first develop and perfect your sales approach with humans: emails, scripts, objection handling, documentation. Then train the AI agent for about a month before scaling. Expect the same rigor as hiring a top-performing SDR, just with AI replacing the first hire. 4. **Segment Ruthlessly:** Never deploy AI across your entire database at once. Instead, batch contacts into groups of 800-1, 000, create sub-agents for different personas (CROs, CMOs, website visitors, churned customers), and assign specific goals (book meetings, sell tickets, re-engage ghosted leads).

Start with low-stakes segments like leads you ghosted or inbound requests you couldn’t fully address. Avoid mission-critical leads early to manage expectations and allow agent ramp-up. 5. **You Need Exactly Two Humans to Succeed:** First, an expert engineer from the vendor side—solution architect or forward-deployed engineer—who helps train and deploy the agent effectively. Without this support, even the best product fails. Second, an internal GTM engineer (“AI nerd”) who manages orchestration, defines CTAs, segments leads, and oversees follow-ups. This person typically isn’t from traditional sales but from marketing, RevOps, or technical roles. Self-serve AI SDR solutions aren’t mature yet; human involvement is essential for training and scaling. **Tech Stack That Works:** We run 20+ AI agents—more than human reps—using Artisan (~6% outbound response), Qualified (~6% inbound response with 130+ meetings since August), and Agentforce (70% open rate on re-engagement). Deployment and tuning take around two weeks, require ongoing review, and connect to a single source of truth for contact assignment. For communication, prioritize chat (preferred by 85% of prospects) before layering voice and video (which add complexity). **Top 5 Mistakes to Avoid:** 1. Keeping humans on tasks they dislike, like ticket follow-ups; AI agents do this better and unlock revenue. 2. Not monitoring every message sent by AI agents early, risking errors and missed training opportunities. 3. Underestimating agent ramp time; proper deployment needs at least two weeks. 4. Building AI strategy around team members who might leave soon, risking disruption. 5. Evaluating too many AI vendors at once, which hampers training and fair assessment; better to select one or three trusted vendors and deepen implementation. **Bottom Line:** By 2026, there’s no excuse for slow human follow-ups. AI SDR products—across chat, voice, and video—are effective but require a strategic approach: replicate human best practices, maintain clear documentation, rigorously train agents, segment contacts carefully, and assign dedicated humans both from vendor and internal teams to oversee rollout. Even a modest 15-20% growth driven by AI agents translates into significant “found” revenue from neglected leads and follow-ups. Your prospects deserve better, and now AI makes delivering that possible.