Open any founder's inbox on a Monday morning. Most of the cold messages start with a line about their "recent post" or their "impressive growth," and you can tell within one sentence that no person read either.
AI did not invent bad outreach. It just made it free to send at scale. The teams winning right now use AI differently: as a research partner and a first-draft writer, with a person firmly in charge of what goes out.
Buyers are tuning out generic AI outreach
Start with the buyer's side, because that is where this gets decided.
According to a Gartner survey of B2B buyers published in June 2025, 73% actively avoid suppliers who send irrelevant outreach. Not "ignore the email." Avoid the supplier. One lazy message can take you off a shortlist you did not even know you were on.
73%
The same Gartner research found that 61% of B2B buyers prefer a buying experience with no sales rep at all. Put those two numbers together and the message is clear. Buyers are doing more on their own, and they are less forgiving of anything that wastes their time.
That does not mean AI is the enemy. It means volume is no longer a strategy. The question is how to use AI to make each message more relevant, not just more frequent.
Where AI genuinely helps
Here is where I have seen AI do real, useful work in sales. These are tasks that are slow for people and fast for machines.
1. Account research. Reading a company's recent news, hiring pages, earnings notes and public posts takes a rep 20 minutes per account. AI can pull that together in seconds and summarize what matters. The rep starts with context, not a blank tab.
2. Prioritization. Most teams work accounts in the order they appear on a list. AI can score accounts against your ideal customer profile and recent buying signals, so reps spend their best hours on the accounts most likely to care right now.
3. Timing. A new leader joins, a company raises money, a competitor gets mentioned, a job post shows a new initiative. These signals are easy to miss by hand. AI can watch for them and flag the moment an account becomes worth contacting.
4. First drafts. Given good research and a clear point of view, AI writes a decent first draft of an email or LinkedIn message. Not a final one. A starting point that a person can sharpen in two minutes instead of writing from scratch in fifteen.
This is not theory. A Gartner survey of more than 1,000 B2B sellers, published in September 2024, found that sellers who partner effectively with AI tools are 3.7 times more likely to meet quota than those who don't.
3.7x
Notice the word "partner." Not "hand everything over to."
Where people must stay in charge
Now the other half. These are the parts of selling where speed matters less than judgment, and where getting it wrong costs far more than getting it slow.
Judgment. AI can tell you a company just hired a new CMO. It cannot reliably tell you whether that CMO is busy putting out fires and would hate a cold email this week. A rep who reads the situation well knows when to wait.
Approval. Every message that goes out carries your brand. A person should read it before it is sent. More on this below, because it is the single most important habit in this post.
Relationships. Trust is built in the back-and-forth: the follow-up question, the honest "we're not the right fit for that," the call where you listen more than you talk. Buyers still want this. Gartner research presented in May 2026 found that 69% of B2B buyers prefer to validate AI-generated insights with a sales rep.
The case for a human approval step
If you take one thing from this post, make it this: no AI-written message goes to a real prospect without a person approving it.
I know the objection. "That slows us down." It does, a little. Here is why it is worth it.
- It catches the bad ones. AI makes confident mistakes. A wrong job title, a stale news item, a tone that is slightly off. A person spots these in seconds.
- It protects your best accounts. Your top 200 target accounts are too valuable to experiment on. One irrelevant message can close a door for a year.
- It keeps your team sharp. Reps who review drafts learn what good looks like. Reps who never look at what is sent lose touch with their own pipeline.
- It keeps volume honest. When every message needs a human yes, nobody is tempted to blast 5,000 emails just because the software can.
The approval step does not have to be painful. The way I like to build it, AI does the research, scores the account and writes a draft with the reasoning attached. The rep reads it, edits a line if needed and approves with one click. That takes a minute or two per message, which is a small price for knowing that everything going out under your name is something you would stand behind.
Gartner predicted in November 2025 that by 2028 AI agents will outnumber human sellers by ten to one, yet fewer than 40% of sellers will say those agents improved their productivity. The gap between those numbers is what happens when teams automate the sending without fixing the thinking.
Let AI prepare the message. Let a person decide whether it deserves to be sent.
A practical split of the work
Here is a simple way to divide sales work between AI and people. Use it as a starting point with your team this week.
Give to AI:
- Pulling account and contact research into one summary
- Scoring accounts against your ideal customer profile
- Watching for buying signals and flagging timing
- Writing first drafts of outreach and follow-ups
- Summarizing call notes and updating records
- Suggesting next steps after a conversation
Keep with people:
- Deciding which accounts are worth contacting this week
- Approving every message before it is sent
- Replying to anyone who responds, especially with questions or objections
- Running discovery calls and live conversations
- Making judgment calls on pricing, fit and timing
- Building long-term relationships with champions and the wider buying group
A quick test for anything in between: if getting it wrong would embarrass you in front of a buyer, a person signs off.
How to start this week
You don't need a big rollout. Pick one step:
- Audit what went out last week. Read 20 sent messages. Would you reply to them? If not, add an approval step now.
- Move research to AI first. Have reps start each account from an AI-prepared summary and measure how much time it frees up.
- Add signals to your prioritization. Even a short list of three or four buying signals will change which accounts get attention.
- Write down your "never send" rules. No fake compliments, no invented familiarity, no message that could go to anyone.
The short version
- Buyers are punishing irrelevant outreach, and Gartner found 73% avoid suppliers who send it.
- Use AI for research, prioritization, timing and first drafts, where speed helps.
- Keep judgment, approval and relationships with people, where trust is won.
- Put a human approval step in front of every message that reaches a prospect.
- Measure success by relevance and conversations, not by how many emails went out.
AI should make your team sound more like themselves, not less. If a buyer can tell a machine sent it, you have already lost the conversation.
