Generative AI could add trillions to the economy, but what does it mean for your sales team? We break down the real-world applications, top tools to evaluate, and how to implement AI without losing your human touch.
We provide neutral advice that works for you.
Away from point sellers, pay to play sites and all the noise that only adds to confusion.
A safe space to ask the questions you want to.
McKinsey reckons generative AI could add between $2.6 and $4.4 trillion a year to the global economy across all use cases. Sales is in that number. So is marketing, finance, customer service, and a lot of slides in board decks. The interesting question is not "is generative AI big," because everyone agrees it is. The interesting question is which of those trillions actually land inside your sales team this quarter, and which ones are still vapour. This post is the practitioner cut. Where genAI pays back in sales today, where it does not yet, and the tools worth your time.
Generative AI for sales is software that creates new sales content (emails, summaries, scripts, talking points, list-builds) on demand, based on a model trained on huge amounts of language data.
It is the AI that writes, summarises, and drafts. ChatGPT, Claude, Gemini, all the big-name models, and the layer of sales tools that sit on top of them. When your CRM auto-summarises a deal, when your engagement platform drafts a personalised opener, when your call recorder spits out a meeting brief, that is generative AI doing the work.
Three things often get bundled, and they are not the same:
Most sales teams need all three. They are not in competition. The trap is buying a "generative AI sales tool" when what you actually needed was better predictive scoring, or vice versa. The function should drive the buy, not the category label.
Generative AI almost never lives as a standalone tool in a healthy stack. It lives as a layer inside the tools you already use. The genAI features inside HubSpot, Salesforce, Rocketphone, Gong, Apollo are usually more useful than a separate "AI sales assistant" you bolt on the side, because they already have the context.
If a vendor is selling you a "generative AI sales platform" that does not connect to your CRM, you are buying a chat interface with extra steps.
The honest map of what works, ranked by how much we trust the output today.
This is the use case everyone wants and the one with the biggest gap between promise and reality. The model can draft a personalised email in two seconds. The problem is the message still sounds like AI on line two. The "too cleanly written" tell is the kiss of death in cold outbound. Prospects spot it, and they have spotted enough of them this year to instinctively bin the next one.
The working version: use generative AI to draft the structure, the bullets, the variants. Have the rep write the actual surface text. AI proposes, human disposes. The teams who skip the human-on-send step end up with a higher send volume and a lower reply rate, which is a worse outcome than sending fewer thoughtful emails.
This is where genAI quietly earns its keep. A model pulls the prospect's recent LinkedIn activity, the company news, the previous call notes, the open deals, and produces a one-page brief that used to take a rep 30 minutes. Reps actually use this. The output is for internal consumption, so the "AI-tell" problem does not apply.
A blend of predictive and generative: the model scores the lead and writes the "why this matters" line. The "why" line is what gets reps to actually call the lead instead of ignoring the score. This works.
Inbound chat that qualifies the lead and books the meeting. Mature. Reliable. Not the same as outbound. Use generative AI on inbound qualifier flows. Do not use it as a substitute for an SDR picking up the phone on outbound.
Auto-summarise the call, auto-update the deal stage, auto-log the next action. Every salesperson, no exception, forgets to update notes at some point. Generative AI fixes this without nagging the rep. The data gets cleaner. The leader's reports actually mean something. Quiet win, real ROI.
The list, organised by job. We have led with the partners and platforms our clients actually deploy.
The most-asked question we get is "which generative AI tool should I buy?" and the most accurate answer is usually "none, you already own one."
If you have HubSpot, turn on Breeze before you look elsewhere.
If you have Salesforce, look at Einstein and Agentforce before you bolt a third tool on.
If you have Gong or Fathom for call recording, use the AI summaries and the AI coaching layer they already ship with.
If you do not have any of those, the right first buy is a call recorder (Fathom is free at the entry tier), not a generative AI platform. The recorder gives you the data that every other genAI tool needs to be useful.
The principle is simple: generative AI value comes from the data it can see. Tools sitting inside your existing data win against tools sitting outside it.
The one decision that matters: does it live inside your CRM, your engagement platform, your call data? Or is it a standalone you have to feed?
A standalone genAI tool needs you to copy context in, paste output back. A native one already has the context. The native one wins 90% of the time, even if the standalone is a marginally better model. The friction of context-switching is what kills sales tool adoption.
Beyond that, the usual list: pricing fit for team size, real customer references in your segment, governance and data residency, a clear opinion from the vendor on what it does NOT do.
Call summaries. Meeting prep. CRM data enrichment. None of these touch a prospect. None of these get a rep into trouble if the model gets it slightly wrong. They give time back. Start there.
If the AI is not measurably moving win rate, cycle length, pipeline velocity, or rep capacity, you are paying for theatre. Pick the KPI before you pick the tool.
A generic model produces generic output. Feed it your ICP, your value props, your real winning email examples, your top customer quotes. Bad input means bad output. The teams that get value from genAI are the teams that put real time into the inputs.
Every prospect-facing message has a human approver. No exceptions. The day you let the AI send to a prospect unreviewed is the day you find out how brittle your model's tone really is.
Two numbers per quarter: how many reps are actually using it (adoption), and what changed on the leading indicators since they started (impact). If adoption is low, you have a training problem. If adoption is high and impact is zero, you have a tool problem.
Generative AI for sales is not the future. It is already in your stack. The HubSpot Copilot is genAI. The Gong summary is genAI. The Apollo email drafter is genAI. The question is not "do I adopt it," it is "do I use what I already pay for, or do I buy a third tool to do the same job worse."
Start with the genAI features inside the tools you already own. If your dialer does not have AI baked in, that is where a fresh buy earns back fastest, a Rocketphone-style AI-native dialer often replaces two or three tools at once. Pick one workflow that gives reps time back. Measure before and after. Then expand. Do not roll out a "generative AI initiative" across the whole team in one go. That is the path to the busy-fool problem at scale, where everyone is using AI to produce more output that nobody actually wants to read.
The unsexy answer keeps being the right one. Get the foundations right, use the AI to remove the boring work, and let your reps spend the saved time on actual conversations.
To give reps their time back. Meeting prep, call summaries, CRM updates, list-builds, all of these used to eat hours a week. GenAI compresses them. Reps spend the saved hours on conversations.
Drafting outreach (with a human on send), summarising calls, prepping for meetings, enriching CRM data, scoring leads with explanations, building prospect lists. The list grows quarterly.
Mostly as a layer inside existing tools (HubSpot, Salesforce, Gong, Apollo, Outreach), occasionally as a standalone assistant. The layer-inside approach wins because it has the context.
For final-mile prospect messaging without human review. For any decision that requires nuance about a specific account or relationship. For replacing the rep on the call. The model is good at structure, less good at judgement.
Because coaching at scale used to be bottlenecked on a good manager's time. GenAI lets you score every call, draft every coaching note, surface every objection trend, without adding headcount. The enablement function is one of the biggest winners of the genAI wave.
👉 AI Agents for Sales: What They Actually Do
👉 AI in B2B Sales
James Donaldson
Founder, Stakki
📧 james@stakki.io
