Blog · AI & Outreach
May 20, 2026 · 8 min read
Outbound used to reward volume. Send enough emails and some would land. That era is over. Inboxes are smarter, buyers are more sceptical, and the cost of a generic message has never been higher. At the same time, the tools available to outbound teams have changed more in the last two years than in the previous ten — and most of that change comes down to large language models.
AI lowered the cost of writing to nearly zero. That sounds like an advantage, and for a few months it was. Then everyone got the same advantage. The result is inboxes flooded with competent, polished, completely forgettable AI copy. When output is free, output stops being the moat. Relevance becomes the moat.
The teams winning at outbound today are not the ones sending the most emails — they are the ones reaching the right person with a message that could only have been written for them. AI helps with that, but only if you point it at the right problem.
Claude is strong at long-context reasoning and at writing copy that sounds like a person rather than a press release. Given messy research — a LinkedIn profile, a press release, a job posting — it distils that into a specific, human observation worth opening an email with. GPT is the versatile workhorse with the widest ecosystem, reliable for structured extraction and fast first drafts at scale. Gemini is strong when the task involves pulling from live search or processing large documents.
The honest truth: the difference between the leading models matters far less than the difference between a good prompt with real research and a lazy one. A frontier model handed nothing but a name and a company will produce confident, fluent, generic slop.
The bigger shift is not the models — it is that AI has moved out of a separate browser tab and into the outbound tools themselves. Clay's AI research agents visit a prospect's website, scan recent news and LinkedIn activity, and write a one-line insight per row — at the scale of a whole list. Lavender sits inside the inbox and scores a draft in real time. Apollo, Smartlead, and Instantly have all folded AI into sequence drafting, subject-line generation, and reply suggestions.
The pattern is consistent: AI is becoming a layer inside every tool rather than a destination you visit. That is a quiet but profound change in how outbound teams spend their time.
AI reliably improves research speed, list enrichment, first drafts, subject-line variation, and reply triage. A team that used to research 30 accounts a day can now research 300 with more depth on each. What does not improve automatically is judgment: knowing who to target, understanding what that buyer genuinely cares about, and having an offer worth their time.
The clearest failure mode is AI slop — emails that are personalised on the surface and empty underneath. "I saw your company operates in industrial manufacturing" is technically personalised and says absolutely nothing. The buyer can feel the difference instantly.
AI raised both the floor and the ceiling of outbound. The floor: anyone can now produce passable copy in seconds — which is exactly why passable copy no longer works. The ceiling: teams that combine sharp targeting, genuine research, and AI leverage can do outreach that was not possible a few years ago.
The teams that win are not the ones that adopted AI fastest. They are the ones that kept a clear strategy and used AI to execute it harder. The tools changed. The fundamentals — right person, real relevance, a reason to reply — did not.