Do AI SDRs actually work?
The early scorecard is poor. Autonomous AI SDRs push more volume into the same collapsing channels, and the category's marquee name, 11x, was reported to have listed customers it did not have and to have inflated its revenue. Even a well-funded challenger, Actively AI, raised $22.5M saying AI SDRs failed by chasing volume. What is working instead is human-in-the-loop: AI finds and drafts, a person sends.
Last reviewed: July 2026
What an AI SDR promises
The pitch is simple: point software at a large list, let it research, write, and send thousands of personalized-looking outbound messages, and book meetings while you sleep. The promise is scale without headcount.
What the results actually show
The category's most-funded name became its cautionary tale. In March 2025, TechCrunch reported that 11x, backed by more than $74M from a16z and Benchmark, had publicly listed companies as customers that had only trialed the product or never used it, drawing a legal threat from ZoomInfo's counsel over deceptive practices. A former employee alleged the company reported roughly $14M in ARR against about $3M in actual contracted revenue and was losing 70 to 80% of customers, while the company claimed 79% retention. Those revenue and churn figures are reported allegations rather than audited numbers, but the pattern, big claims outrunning real results, is the story of the category so far.
Why more volume is the wrong lever now
AI SDRs pour more messages into channels that are already breaking. Cold email replies sit under 1% in large datasets (Belkins, 2025), 73% of buyers avoid vendors who send irrelevant outreach (Gartner, 2024), and the 2024 deliverability rules punish bulk senders. Automating the old volume playbook mostly automates the part that stopped working, faster, and can burn your domain and brand in the process.
The human-in-the-loop alternative
The counter-movement is funded too. In April 2025, Actively AI raised $22.5M explicitly arguing that classic AI sales tools failed by chasing pure volume instead of targeting the right accounts. The durable model keeps a person on the send: let AI do the heavy lifting it is genuinely good at, finding the right people and drafting a relevant first touch, and let a human decide what actually goes out. That keeps outreach relevant and protects your reputation.
Slingapult's read: we are not anti-AI. We are anti-autopilot. Slingapult uses AI to find the warm people and draft the message, then puts you one click from sending. Volume was never the constraint. Relevance was.