An AI SDR agent researches prospects, writes and sends outreach, handles replies and books meetings with little human involvement. It is one of the fastest-growing and most contested categories in sales AI.
Where agents tend to work
- Inbound leads humans don't reach: low-score or after-hours leads that would otherwise get no follow-up. Salesforce said only one in four of its leads were historically followed up; its own agent then re-engaged tens of thousands of leads and booked 800 meetings in its first months.
- Dormant warm contacts: past event attendees, old trials, closed-lost deals. In a Salesforce customer story, SaaStr used an agent on nearly 1,000 untouched warm prospects and reported a response rate above 10% and $2.7 million more revenue closed.
- Structured qualification: asking standard questions, routing qualified leads to the right rep fast.
Where they struggle
- Cold outbound to senior buyers, where relevance and credibility matter most and a generic message damages your name.
- Complex products where the agent cannot answer follow-up questions well.
- Poor data: wrong contacts, no suppression lists, no view of open deals.
The market is noisy
In March 2025 TechCrunch reported that AI SDR startup 11x had shown logos of companies that said they were not customers; ZoomInfo said it had only run a one-month trial in which the product "performed significantly worse than our SDR employees". Former employees alleged high churn, which 11x disputed. The lesson is not about one vendor: verify references and run your own trial with a comparison group.
Measure quality, not volume
Agents make sending nearly free. Judge them by positive replies, meetings held, meeting-to-opportunity rate, pipeline per dollar and unsubscribe and spam-complaint rates — and compare with human SDRs on the same segment. The SDR economics tool in this hub lays out the comparison.