Is your investment in AI reaching key client relationships?

And why the time it saves should go to growing the clients you already have

Most of what is written about AI and selling is about winning new buyers. This piece is about something closer to Symbioss and its focus: what AI is doing to the client relationships you already have and how you manage them best, so you benefit from the growth those enhanced relationships should produce.

The question your clients are starting to ask

Earlier this year KPMG International told its own auditor, Grant Thornton UK, to pass on the savings it was making from AI. KPMG said it would look for a new auditor if the audit fee didn’t come down. The fee fell by about 14% (Financial Times, Feb 2026). If one of the Big Four will have that conversation with a supplier, you can expect your clients to have it with you.

But are clients actually seeing those savings? In a survey of 657 in-house lawyers across 30 countries, 59% said they had seen no noticeable savings yet from their law practices’ use of AI (Association of Corporate Counsel and Everlaw, Oct 2025). We have all used AI, so we know the time it can save, in theory. So what is happening to those savings?

When the Agency Management Institute asked 400 client decision-makers what they expected from their agencies because of AI, only 29% said lower fees, while 52% wanted better strategy and 52% better ideas. And 82% said transparency about use of AI increases their trust (Agency Management Institute, 2026).

In our view, how this conversation proceeds depends on the client relationship. Where it is strong and your value is understood in the client’s own terms, AI is a chance to show how you can serve the client even better. Where it isn’t strong, the discussion risks shrinking to price, and the incumbent is at greater risk, not least because it is negotiating from the weaker side of the table, but also because AI helps the client to be better informed.

If one of the Big Four will have that conversation with a supplier, you can expect your clients to have it with you.

Judged before anyone picks up the phone

Procurement teams are using AI too. In Ironclad’s latest survey, 92% of 800 procurement professionals use it, with contract analysis (39%) and finding and evaluating suppliers (37%) among the most common uses (Ironclad, Oct 2026). We have seen no evidence that AI is choosing suppliers. But your contract, your performance and your rates are increasingly being reviewed through a tool before anyone speaks to you.

This matters most at renewal. Value that sits in a partner’s head, however real, can’t be seen by a procurement system, and goodwill doesn’t show up in a search. Much of what you have achieved for a client is confidential, so it won’t be anywhere an AI tool can find it either. An incumbent facing a re-tender therefore needs its own documented evidence of what it has delivered, shared as a matter of routine and in the client’s numbers, ready to put in front of the people making the decision. And more than ever, a systematic approach to each client relationship, rather than individual heroics, makes it easier for others to understand it.

An incumbent facing a re-tender needs its own documented evidence of what it has delivered.

When every supplier starts to sound the same

AI is good at drafting emails, meeting notes and proposals, and that brings its own problems. In an experiment at UCL and Exeter, stories written with AI ideas were rated better, but they were also more alike (Doshi and Hauser, 2024). A separate study of 319 knowledge workers found that the more people trusted AI, the less critically they thought about what it produced (Lee and colleagues, 2025).

It’s further compounded if polished emails that could have come from any of your competitors are routinely created and sent. Junior people who always start from a draft never build the judgement your best people have, and they miss out on this more subtle thinking, which develops risk across your whole business. Knowing that a finance director is under pressure, or that this is the wrong week to raise a renewal, matters even more when anyone can produce a polished email. So leaders need to protect that judgement and make sure it is passed on to less experienced colleagues.

On trust, our view is that AI’s main contribution is removing the reasons for distrust: the missed follow-up, two people on your team giving different answers, the history that left with a colleague. Trust itself is still earned by people, but AI can increase levels of trust. However, there is a balance here.

The danger comes when AI starts running the relationship rather than supporting it. If drafts go out unread, meetings are summarised but never discussed and nobody senior has their finger on the pulse, the organisation has handed its judgement and its involvement with the client to a tool. Once that sense of the relationship starts to slip, the drift can become inexorable. That is a real risk, because the client may notice before you do.

Spend the saved time on growth

Gartner found AI saves sellers an average of 4.8 hours a week, yet 72% of sales organisations report low reinvestment of that time into high-value work. Among organisations making moderate to large savings, those that do reinvest are 2.2 times more likely to exceed their customer growth goals (Gartner, May 2026).

So the leadership choice is where that time goes. It can go to service cost, with fewer people looking after each client, or to growth yield, with more senior time on the clients with the most potential. In our experience, existing clients are where most mid-sized services organisations find their growth, so thinning out those relationships to save cost puts the numbers at risk. And if nobody decides, Gartner’s figures suggest most of it won’t be reinvested at all, which is a worry when the AI tools and implementation of them cost real money.

Thinning out your existing client relationships to save cost puts the numbers at risk.

Making that choice well depends on understanding each relationship properly: where it is heading, how strong it is and how much potential it has. That needs good data, so the time goes where it will make a difference and not into a relationship where the client isn’t going anywhere.

Governance

All of this needs a few written ground rules, owned by someone senior. Two are easy to overlook. The first is that a person stays accountable. In the Agency Management Institute research, 89% of clients said agencies must remain fully accountable for their recommendations, even when AI is used, so the client lead, not the tool, decides what the relationship needs next. The second is your own voice. AI tends to make everyone’s writing sound alike, so build a clear house style (your tone, the words you use and avoid and how you present evidence) into the instructions your AI tools work from. That way, everything you send a client still sounds like you.

What we would do now

01. Talk to your key clients about AI, and about value, before they raise it. Say what you use AI for, what it changes for them and where the saving goes: more time with them, sharper thinking or, where it’s fair, a lower fee. And share evidence of what you have delivered as a matter of routine, in their numbers, so it is already in front of the people deciding when renewal comes round.

02. Keep the judgement with your people. Make sure each client lead stays close enough to know what makes the relationship tick: reading what goes out, discussing what the meeting notes say rather than just filing them, and spending real time with the client. Pair less experienced colleagues with your best people on key clients, so their judgement is passed on.

03. Get your client data right. AI is only as good as the data you give it, and it draws confident conclusions from poor data. Start with one accurate client list, correct revenue by client, current contacts on both sides, the commitments you’ve made, what each client values and the early signs of drift. Then segment on potential, not just this year’s revenue: a small client today may be a large one in three years.

04. Decide where the saved time goes, and measure the return. AI tools cost money to buy and to implement, and time saved that never reaches a client doesn’t pay that back. Because the gain is often better work rather than fewer hours, agree a handful of measures up front, such as senior time spent with key clients, how many of the client’s people you know well, what clients say about the quality of your advice and growth from existing clients.

05. Write your AI rules down. Include who checks anything AI produces before it reaches a client; what client information may go into which AI tools, and what your client contracts say about it (among UK staff in one survey of large organisations, 23% had put customer data into public AI tools: PagerDuty, June 2026); how open you will be with clients about your use of AI, given that 82% say transparency increases their trust; and a pause to think before anything important goes out. Then review the rules regularly, because the tools and your clients’ expectations will keep changing.

It’s far better to have that conversation on your terms than to wait for a client to start it. And in that spirit: this blog was written with help from AI (Claude). Did it save time? Yes, but it still took us hours to get right. AI helped us organise our thinking and sharpen the argument. The judgement, the evidence checks and the final words are ours. If you’d like to work out what your answer should be for your key clients, get in touch.

About the author

Liz Ashton is the founder and director of Symbioss. We help UK B2B services organisations manage their client relationships as valuable assets, through tailored reviews and programmes built on the RelaValue System. A more systematic approach helps you protect revenue, reduce relationship risk and create more reliable growth. To talk about what it could mean for you and your key clients, get in touch at liz@symbioss.co.uk or www.symbioss.co.uk.

Photographs: Vitaly Gariev, Olena Kholina and Annie Spratt / Unsplash

Sources: Financial Times, KPMG pressed its auditor to pass on AI cost savings (6 February 2026); Association of Corporate Counsel and Everlaw, Generative AI’s Growing Strategic Value for Corporate Law Departments (2025); Agency Management Institute and Audience Audit, Agency Edge 2026 (2026); Ironclad, State of AI in Procurement (2026); Doshi & Hauser, Generative AI enhances individual creativity but reduces the collective diversity of novel content, Science Advances 10(28) (2024); Lee et al., The Impact of Generative AI on Critical Thinking, CHI 2025 (2025); Gartner, survey of 210 chief sales officers and senior sales leaders (19 May 2026); PagerDuty and Wakefield Research, Shadow AI Survey, UK report (2026).