Observe ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- AI brokers can cut back service prices by greater than 20%.
- Service leaders are investing in tech and information integration.
- Solely 46% of service reps’ time is spent participating prospects.
The 2025 State of Service report from Salesforce discovered that 4 out of each 5 service leaders say AI agent investment is crucial to assembly enterprise calls for.
The report highlighted a number of key findings, together with adoption charges of AI brokers, enterprise alternatives and challenges, and, most notably, how firms are incorporating predictive, generative, and agentic AI to ship sooner, extra correct, and extra customized interactions. Leaders count on AI brokers to spice up prior AI-enabled outcomes and are backing that expectation with funding.
Additionally: Only 11% of business leaders see AI leading to major job cuts – for now
To raised perceive the impression of AI on the client help and repair trade, together with the adoption of agentic AI within the enterprise, I interviewed Michael Maoz, senior vice chairman of innovation technique at Salesforce, one of many world’s main specialists on buyer relationship administration (CRM), buyer expertise (CX), data administration (KM), and customer support.
Earlier than becoming a member of Salesforce, Maoz was a analysis vice chairman and distinguished analyst at Gartner, serving because the analysis chief for the customer support and help methods space. Listed below are the highlights from my dialog with Maoz.
1. AI brokers are taking maintain
We’re seeing the rise of autonomous methods that sense, cause, determine, and act, all with minimal human enter, and it might be attention-grabbing to see how that development, which ideally needs to be guiding a enterprise’ architectural playbook, is enjoying out within the new Salesforce State of Service report. What did you discover?
Michael Maoz: The customer support leaders that we interviewed are each visionary and sensible. They may not be prepared for autonomous methods, however they clearly talked about their intention to ramp up the usage of agentic AI. I discovered 4 highlights within the report:
- Generative and agentic AI take away what’s boring and repetitive, and as a substitute create the area for buyer reps to deal with extra complicated duties.
- Correct information out there for the time being of buyer engagement raises buyer satisfaction and lowers prices.
- Higher service course of design, with nice data administration methods and real-time insights, upskill buyer representatives as by no means earlier than.
- The way forward for buyer engagement is multimodal.
The final one, multimodal, was a shock. There may be a whole lot of pleasure round including video, photographs, and sensor information (collectively referred to as multimodal) to service responses.
2. Client-facing companies cleared the path
In all my conversations with our prospects and companions this 12 months, I’ve heard that the usage of AI brokers is taking off. That’s very true in consumer industries. These sectors embrace monetary providers, journey, and retail. I have been displaying individuals how Salesforce is experiencing 85% autonomous decision of service instances, and the lesson there’s that highly effective AI and entry to data, tied to understanding the client context, results in success.
Additionally: Nearly everything you’ve heard about AI and job cuts is wrong – here’s why
Service will likely be half-automated by AI inside two years. The Agentic Enterprise Index’s service outlook is blunt: by 2027, roughly 50% of service instances will likely be resolved by AI, with most service leaders calling AI-agent funding “important” to assembly demand. The thrust is price, pace, and buyer satisfaction positive factors from end-to-end agent workflows.
Michael Maoz: As I went by the survey, I assumed again to a robust picture you share of kids ingesting water from a faucet. Your level is that information is like water: it does nobody any good until it’s clear and accessible. In our survey, which reached 6,500 service professionals across the globe, we requested respondents what they see as most crucial parts for accelerating AI adoption. The reply got here again to precisely your level: related, trusted information is the bedrock of profitable AI adoption.
A robust information basis is crucial for driving enterprise success, and it must be discoverable and accessible. Whether or not the dialogue is about learn how to upskill reps so they’re extra intuitive and higher at their jobs, or how AI can create actions, the trail to realizing these capabilities rests on correct, trusted information. Information comes from structured databases and unstructured content material in paperwork, photographs, and sensor indicators.
3. Customer support is a crew sport
Manner earlier than the introduction of gen AI and agentic AI, corresponding to Salesforce’s Agentforce, companies understood {that a} single view of the client was a differentiator. The problem is that bringing collectively correct information is essential to totally realizing the advantages of AI. For a customer support group, there are a number of challenges, not the least of which is that they’re supporting an growing variety of areas, corresponding to advertising and marketing, e-commerce, and social media websites, in addition to their conventional roles.
Prospects really feel one thing virtually magical when a service group has all the related data for the interplay at their fingertips (even when these are the digital fingertips of a chatbot), persistently throughout the engagement channels.
Michael Maoz: Positively. From the survey, your level a couple of single view of the client come outs. For our prospects who’ve unified their information, Agentforce takes present workflows, and all the present actions, and creates autonomous actions with them.
Additionally: Your coworkers are sick of your AI workslop
Others use the a whole lot of prebuilt Agentforce workflows, they usually additionally construct their very own. The outcomes are superb. That explains our findings within the seventh State of Service report that organizations that unify service channel information in a single platform usually tend to name their AI very profitable.
4. Information silos have to be addressed
We typically neglect that behind the scenes, the common enterprise gathers and maintains between 20 and 200 separate information factors a couple of buyer. It’s tough for many organizations to drag all the proper information factors collectively in actual time as a result of it’s scattered throughout impartial silos.
Credit score data is held within the monetary system, order data is within the ERP or commerce system, and buyer channel habits is held by advertising and marketing, as is loyalty and membership information. In consequence, the rep should soar from system to system, piecing the data collectively.
Additionally: Your next job? Managing a fleet of AI agents
I discovered an incredible statistic within the survey. The highest-performing service organizations have discovered the important thing to overcoming the problem of dispersed information is to prioritize information integration. The survey says that 88% of leaders are prioritizing tech integration. And 44% say information silos have already delayed AI tasks.
Michael Maoz: We proceed to see voice calls with human help representatives dealing with as much as 35% of all customer support wants. Once we dig in, prospects name a enterprise out of frustration and for lack of a greater different, not out of selection. The explanation for these calls is frustration over the failure or unreliability of the opposite choices, corresponding to chat and chatbots.
That 35% of telephone calls additionally masks the big strides in self-service that almost all prospects use with out serious about channel selection as a result of it feels intuitive and pure. Take a look at banks, on-line retailers, and airways. They’ve shifted 90% of service interactions to cell wallets and apps, kiosks, and web sites. The identical is true in lots of different industries, corresponding to media and telecommunications.
5. Service execs can deal with worth
Self-service has undoubtedly improved. On the identical time, most organizations have a protracted method to go. The following degree of automation and self-service is agentic AI, corresponding to Agentforce. The survey confirmed that 79% of leaders agree that AI agent funding is crucial.
Additionally: AI helps strong dev teams and hurts weak ones, according to Google’s 2025 DORA report
Agentforce strikes past producing the correct content material to performing actions beforehand carried out by people. It’s extra vital than ever to assist service representatives spend much less time on repetitive, boring, however required actions. On the identical time, since self-service has solved the simplest issues and left representatives with tougher work, giving people a lot better instruments to satisfy the brand new necessities turns into actually essential.
Michael Maoz: That brings us to upskilling and caring for the client consultant. They see AI coming. In addition they report that, in 2025, solely 46% of the rep’s time is spent participating prospects. That 46% of the time is upleveled when wealthy, customized, contextual information and insights can be found.
What in regards to the different 54% of consultant time? That’s spent on case notes, assembly notes, and trying to find data. They’re thinking about optimizing all of that information, and a good way is utilizing Service Cloud instruments, corresponding to Enterprise Data and Agentforce. The time saved may be redirected to the advertising and marketing and gross sales journey.
6. AI brokers take away wasteful actions
Service leaders usually ask me the place to start. My commentary is that main service organizations stock probably the most logical duties for Agentforce, corresponding to refunds, rescheduling, or dispatching a technician. Typically, FAQs are the plain place to start out, transferring to dialog evaluation and abstract, data discovery, after which onto actions, corresponding to appointment modifications, order returns, kind completion, managing the return of broken objects, and a whole lot extra.
Additionally: OpenAI tested GPT-5, Claude, and Gemini on real-world tasks – the results were surprising
A greatest apply we see is to arrange a dashboard corresponding to Tableau Pulse to see vital information, such because the change in difficulty decision charges, common dealing with time (AHT), and buyer satisfaction. The survey exhibits that reps can notice 20% cuts in service price, and far of that discount is as a result of the decision instances for his or her representatives enhance.
Michael Maoz: It’s typically a enjoyable train to ask a enterprise if they’ve requested prospects the query: ‘Have you ever switched to a brand new firm due to a nasty advert placement or advertising and marketing message?’ It is unlikely the reply is sure. Customer support budgets needs to be a lot larger. Most buyer surveys rightly deal with the impression of unhealthy customer support, the place 43% of shoppers say that they’ll minimize spending on account of unhealthy service.
Customer support sits on the intersection of want and emotion. One thing has gone incorrect; there’s a drawback, difficulty, or uncertainty. Checked out as a possibility, when a enterprise anticipates an rising difficulty or instantly resolves a buyer drawback, the client is calm and probably receptive to a advertising and marketing message or a gross sales provide. Gross sales and advertising and marketing organizations are more and more conscious of the big alternative to develop buyer spending by offering a differentiated service expertise.
This consciousness would possibly clarify the statistic that 70% of service leaders count on to obtain bigger budgets to carry out advertising and marketing and promoting duties, or to coach a brand new technology of representatives to work with superior AI. The query enterprise leaders should ask is, ‘How will we get advertising and marketing and customer support to work higher collectively?’ We’re completely seeing the reply at visionary firms. Within the survey, 81% of service representatives mentioned constructing relationships with prospects is a crucial a part of their job.
Additionally: Jobs for young developers are dwindling, thanks to AI
In my dialog with Maoz, he jogged my memory of an essential takeaway: “From 2025 to 2030, practically each service group will face this query: How will we help our individuals in a world more and more powered by AI? There isn’t any single reply, however there’s a north star. The businesses that succeed will not be those that automate the quickest. They will be those that lead with function, talk with empathy, and design with humanity in thoughts.”
Maoz highlighted key findings of the State of Service report that showcase the significance of tradition, individuals, course of, technique, and lasting know-how — maybe, ordered precisely, as key enterprise success components in an AI-powered financial system.
The seventh version of the State of Service report is crammed with a whole lot of vital insights. Agentic AI removes repetitive duties and opens up time for the human rep to be extra inventive. Clear information and contextual data are in excessive demand. Consistency throughout channels is crucial to buyer loyalty. The way forward for buyer engagement is multimodal.
Michael Maoz will likely be protecting the State of Service report extensively at Dreamforce. The session particulars are: DF25 5268 – Unlock the Future: People & Brokers Unite to Redefine Service, 12:30 PM on 14 October. Hosted by Maoz, the session will embrace a panel dialogue with the most effective and brightest service executives and trailblazers from varied industries.
This text was co-authored by Michael Maoz, senior vice chairman of innovation technique at Salesforce.