AI should not just make the salesperson faster. It should make the sales organisation better.

In Article 3, I discussed modern GTM as the company's overall commercial architecture. In this article, I examine how AI can improve execution — and why automation is not necessarily the same as improved commercial performance.

A salesperson can now research a prospect in a few minutes. AI can analyse the company, identify relevant contacts, draft outreach suggestions, summarise meetings, update the CRM and suggest the next action. It sounds like a significant productivity improvement. And it can be.

But imagine the salesperson uses the time freed up to contact even more customers who should never have been in the target group. Then the technology has made the process faster. It has not necessarily made the sales organisation better.

AI can accelerate good commercial execution. It can also accelerate poor commercial execution. In my view, that is one of the most important leadership challenges in modern sales.

Productivity is not automatically growth

Gartner's survey from May 2026 showed that AI tools freed up an average of 4.8 hours a week for salespeople in the organisations studied. At the same time, 72% of sales organisations reported a low degree of reinvesting the time saved in more value-creating activities.

That raises a question that should be central to any AI investment: what should the employee use the freed-up time for? If the answer is simply more activity, the company risks confusing efficiency with effect.

Instead, it should be a matter of better market prioritisation, more qualified customer conversations, stronger account development and higher quality in the commercial decision process.

AI should be used where it creates real value

McKinsey's July 2026 report on the future of B2B sales argues for rethinking connected commercial workflows rather than simply implementing individual AI tools. That is a significant difference.

An AI assistant that drafts an email can save time. A connected system that helps the company identify attractive customers, detect relevant buying signals, prioritise the sales effort, support account plans and follow up on value realisation can affect a far larger part of the commercial model. The precondition, of course, is that data, processes and decision ownership fit together.

Where should the people be?

I would start by distinguishing between three types of commercial work.

Work that can usefully be automated

For example data collection, meeting notes, routine CRM updates, information searches and first drafts of documents.

Work where AI should assist a person

For example prioritising customers, account analysis, developing value propositions, qualifying opportunities, preparing proposals and assessing commercial risks. Here the technology can provide suggestions, connections and relevant information, but the conclusion must still be quality-assured.

Work where human involvement should play a central role

This applies particularly to complex customer conversations, negotiations, handling conflicts of interest, strategic relationships and situations where the company's credibility or significant customer promises are at stake.

The boundaries will shift with the technology. They will also depend on the company's industry, customer type and risk profile. The point is therefore not to lock in a universal division of labour. It is to make deliberate choices about what should be automated, what should be supported and where people should keep responsibility.

The big organisational risk

Salesforce's State of Sales, 7th Edition from 2026 shows widespread interest in AI agents among the sales organisations surveyed. Bain's 2025 Commercial Excellence research also indicates that companies creating more value through AI work with, among other things, a stronger technology foundation, better integration and more systematic implementation.

That underlines a problem I would pay particular attention to: companies can very easily end up buying more tools without improving the way the work is done.

If customer data is unreliable, the ICP is unclear, account ownership is fragmented and the sales process does not reflect the customer's buying journey, AI becomes yet another layer on top of a weak commercial model. Technology alone cannot resolve the leadership decisions the company has failed to make.

Start with the commercial constraint

Before recommending a major AI investment in sales, I would ask four questions:

1. Which concrete commercial problem are we trying to solve?
2. Which decisions or workflows are preventing us from creating better results?
3. How should the work be divided between people and technology?
4. How do we document the effect on quality, conversion, growth, customer value or costs — not just on time spent?

It is necessary to be able to measure both gains and unintended consequences. A sharp increase in the number of outreach messages sent can, for example, be a poor result if relevance falls at the same time.

A more human sales organisation?

Perhaps the most interesting thing about AI is not how many salespeople the technology could potentially replace. It is the opportunity to free skilled people from work that does not use their real capabilities. Less administration. Better preparation. More time to understand the customer's business. Better support for commercial decisions.

But that does not happen automatically. It requires management to rethink the work and actively decide where the freed-up capacity should create value.

The question is not how much sales work we can get AI to do. The question is how people and technology together can create a stronger commercial organisation.

Next article: The sales organisation of the future must be built around customer value, not traditional silos. Because when buying behaviour changes, GTM becomes more connected and AI takes over new tasks, it becomes necessary to reconsider roles, leadership and how commercial work is organised.

The sources cover technology use, organisational implementation and the link between time savings and commercial results. The division of work between automation, AI assistance and human responsibility is a recommended leadership model, not a universal research conclusion.