The first time I was targeted by a fully automated outbound campaign, I was genuinely curious. The first message gave the framework away. Work long enough with ChatGPT or Claude or any other LLM, and you learn the models' preferences for certain framings and words. But who cares, really. The message was well crafted and on topic. The demo was a success. I was ready to buy.
Then I did something the system had not planned for. Instead of replying on LinkedIn, I emailed the CEO directly, from an address the system did not know. The conversation stalled. We ended up in a loop, repeating information we had already agreed on. Then the second failure. Because nothing had “progressed” on LinkedIn, the bot started sending me warm-up messages on that channel. To a buyer who was ready to sign.
Small inconsistencies can be forgiven. A useful tool is a useful tool. What you cannot forgive is a deal that stalls and disappears at the exact moment the buyer wants to move. And there is a quieter cost. Think about everything a good seller learns in the conversations that lead to a close: objections, priorities, the buyer's own language, timing signals. All of that context is lost while the bot hammers away on autopilot, unaware there is a bigger picture.
Automation is excellent at the repeatable parts of outbound. Buying decisions are not one of them. AI makes execution cheaper. It does not yet make judgment optional.