AI: Powering Up Your Sales Game
When I wrote this in February 2024, AI in sales was something only some teams used. That has flipped. Salesforce's 2026 State of Sales, a survey of 4,050 sales professionals carried out in August and September 2025, found that 87% of sales organisations now use some form of AI, and 54% of salespeople say they have used AI agents.
So the useful question now is where AI helps, and where it makes things worse.
Where AI earns its keep
1. Research before the first message
Reading a prospect's profile, recent posts and company news takes time for every single person. Salespeople in the Salesforce survey expect agents to cut prospect research time by 34% and email drafting by 36% once fully in place.
2. Writing for one person
A template has to work for everyone, so it says nothing specific to anyone. We tested this for one client over 16 weeks, with the same audience and the same sequence. Messages written for each person drew replies from 13.6% of people, against 7.7% for the client's own templates. Here is how that test ran. A reply means any reply, including "no thanks".
3. Call notes and the CRM
AI note-takers are now built into many sales tools. HubSpot, for example, lists a notetaker and a prospecting agent from its Sales Hub Starter tier upwards. The win for a small firm is simple: the call summary and next step land in the client record without anyone typing them.
4. Forecasting and pipeline
This is the enterprise end of the market, and it has consolidated. Clari, which I mentioned in 2024 as a forecasting tool, completed a merger with Salesloft in December 2025. If you are a team of one to ten, a clean CRM and a weekly report matter far more than a forecasting platform.
Where AI goes wrong
More AI does not automatically mean better results. Gartner predicts that by 2028 AI agents will outnumber salespeople ten to one, yet fewer than 40% of salespeople will say agents improved their productivity. Its warning is that agents scale whatever system they sit in, including a messy one. Salesforce found the same thing from another angle: 51% of sales leaders using AI say disconnected systems are slowing their AI plans down.
Left unchecked, AI also writes badly in predictable ways. It invents a shared connection, guesses at dates, or reaches for the same stock phrases as every other automated message. That is why every draft we send passes a quality gate that checks it against the client's own rules. A draft that fails gets three rewrites; if it still fails, that draft is held and never sent, and the message is written again from scratch the next day. So far that gate has stopped 1,420 weak drafts before anyone saw them.
What it costs
Less than most people expect. In August 2026, the AI cost of writing and checking our outreach messages came to about $3 to $6 per 1,000 messages. The real work is the setup: a clear ideal client, a clean list and written rules for what a good message looks like.
Where to start
- Write down who your ideal client is, specifically enough that two people would build the same list from it.
- Clean your CRM before you add any AI to it.
- Use AI for research and first drafts, and keep a check between the draft and the send.
If you would rather not build this yourself, it is exactly what we run for clients through done-for-you outreach on LinkedIn and email.
