AI will have a huge impact on Customer Experience in 2025—that much everyone agrees on. But how and why? We’ve seen lots of hype around 2025 trends, so we want to share our realistic assessment of where and when AI will have the most impact across marketing, sales, and support. In this article, we call out our top list of AI trends and, most importantly, what you should do about them in 2025.
When we describe AI use cases, we use the following graphics to indicate how often we have seen them deployed in corporations today, in pilot or full production.
Our Pick of The Top CX AI Trends to Act On Now
-
At-scale Hyper-personalization
-
AI-enhanced Sales Reps and Support Agents
-
AI Coaching
-
AI Assistants for Sales and Support
-
AI-enhanced CX managers
1. At-scale Hyper-personalization
Hyper-personalization allows customer-facing organizations to create individually-tailored customer experiences, using data analytics, predictive analytics, data analytics, machine learning and LLMs to tailor content and experiences to individuals based on their unique interests, preferences, behaviors and demographics. It goes beyond simply addressing someone by name or recommending products based on past purchases — it’s about crafting a bespoke experience for each user, just like a human sales or support professional would.
How mature is it?
Hyper-personalization is starting to be used, successfully, for marketing and sales. It’s almost unheard of in support at this point, but we expect the same AI techniques to work very well and take support interactions to the next level.
How can I use it?
Here’s a list of use cases to consider, divided between Sales and Marketing and Support. We rate each use case by likelihood, with the most well-established use cases at the top.
Sales & Marketing
- Personalization of messaging and content based on the characteristics of the company (industry, size, geo, competitive offerings used etc.) and the contact (title, background, tenure at the company, familiarity with the offering).
- Playbook-based personalization based on first, second and third-party buyer intent data about the buyer (e.g. challenges, pain points, product interest).
- Personalization based on buying stage, taking in consideration the position of the prospect in the buying journey (awareness, research, decision, etc.)
- Personalization that incorporates historical communications, resulting in very targeted sales and marketing communications and offerings for the prospect. Style of communication can also be adjusted based on cultural, generational, and personal style of prospect, and what worked or didn’t work in the past.
- Account-based or Buying Center personalization that takes into account the multiple contacts involved in the buying decision.
Support
- Personalization based on the products or services the customer purchased.
- Personalization based on the level of knowledge of the customer (how long they have used the product; how technical they are), their media preferences (e.g. videos vs technical paper), and their communication preferences.
- Personalization based on the specific use cases the customer is addressing.
2. AI-enhanced Sales Reps and Support Agents
What is it?
AI assistants can help human reps and agents do their work faster and better. Think of them as giving your team members superpowers, like having a personal, talented assistant by their side 24×7. AI assistants can perform research, deliver summaries, draft communications, prioritize prospects based on intent and customers based on sentiment analysis, and suggest actions across the customer experience journey.
AI assistants go beyond productivity improvements and can help transform the Customer Experience at scale with personalization techniques that were previously possible only for top-tier customers and prospects.
While the use cases are very powerful and exciting, what you will be able to achieve depends, a lot, on your market and data. For example, it is easy for AI assistants to research and prepare for a call with a public company, but if your organization is focused on SMBs there will be much less useful public data available to the AI assistant (or the sales rep, or the CSM!). Similarly, if your internal knowledge base is weak, don’t expect great results from AI: pointing a search tool to your case data is unlikely to deliver the targeted results you would want.
How mature is it?
Compared to other AI trends, AI-enhanced agent technology is pretty mature, with solid commercial tools available that handle most of the simpler use cases. Advanced use cases are still a bit futuristic. You’ll want to test them against your own situation, of course, but we are hearing plenty of success stories.
How can I use it?
Sales & Marketing
- Sales reps can prepare for meetings with prospects in no time by getting a concise summary of their industry, company, and role. AI assistants gather information from internal sources (CRM) and external sources (10K filings, LinkedIn, and the corporate website).
- Sales reps and CSMs can use AI to draft meeting follow-ups, complete with summary of the meeting and action items; create custom presentation decks; and answer RFPs.
- AI assistants can recommend actions based on best practices and success from other sales situations: discuss pricing and prepare quotes, suggest a case study or white paper, or involve new contacts in the sales process.
- AI assistants can help sales reps and CSMs by performing data entry tasks: update CRM records based on notes and voice commands, or add new contacts based on recent meetings or new employees and promotions.
- AI assistants can autonomously identify relevant news and changes in target accounts, and help re-prioritize accounts activities using specific sales playbooks.
Support
- Support agents can be alerted when customers are getting frustrated or are more likely to escalate, allowing them to prioritize their work.
- AI-enhanced search tools enable support agents to find answers to customer queries without having to reinvent the wheel–and, if needed, easily collaborate with other agents who are working on similar issues.
- Support agents can use GenAI to draft responses to customers. For example, instead of using a standard KB article to respond to an inquiry, AI agents can customize a step-by-step resolution guide for the customer situation.
- GenAI can draft very serviceable knowledge base documents.
- Support agents can get a summary of customers’ relative value to the organization, their use cases, and even how technically knowledgeable their contacts are.
3. AI Coaching
What is it?
AI coaching can provide real-time feedback during or right after live prospect and customer interactions, using speech and sentiment analysis to evaluate communication skills, tone, and effectiveness in building rapport with prospects and customers. It can also automate onboarding with role-playing and simulations for practice in a risk-free environment.
How mature is it?
Full AI-enabled coaching is still a ways away, but some sentiment-analysis tools make solid suggestions for conducting fruitful prospect and customer conversations. Some tools are starting to do a decent job evaluating agents’ work in particular language quality and adherence to process.
How can I use it?
Sales & Marketing
- Sales reps and CSMs can get instant feedback on how well they handle customer conversations during live meeting, using sales enablement tools or AI extension of web meeting technology.
- GenAI can critique and make improvements suggestions for written communications, including emails, proposals, presentations, etc.
- AI avatars based on specific ideal customer personas (ICPs) can help new sales reps practice their sales skills in a risk-free, realistic environment.
Support
- Propose in-the-moment corrections and improvements to customer communications.
- Complete case quality checklists (independently for language and process, but not for depth of troubleshooting or any other sophisticated behaviors).
4. AI Assistants for Sales and Support
What is it?
AI agents are autonomous robots that can perceive their environment and context, process information, make decisions and take actions to achieve specific goals without human intervention. How mature is it?
AI agents are still a work in progress but progress is fast and impressive. Today, AI agents handle basic interactions very well. We expect that the bots will be getting better at interpreting and responding to emotions, above and beyond factual content.
How can I use it?
Sales & Marketing
- Sales bots can answer factual questions about the product and its use cases and benefits for non-enterprise products. (Note that if a human would often answer “it depends” to a prospect’s question, your product is probably too configurable and customizable for a self-sufficient sales bot to handle the prospect.)
- Handle simple phone interactions in multiple languages: answer product and pricing questions from prospects, book meetings with sales rep.
- AI assistants can develop new leads by autonomously reaching out to contacts with personalized communications (e.g. via email or LinkedIn messages), interacting with prospects and setting up appointments for sales reps.
- Sales bots can handle full sales conversations via voice, text or chat, through the entire process of understanding needs, recommending products, negotiating pricing, and completing an order.
Support
- Chat service bots work well with reasonably circumscribed use cases such as giving order status, processing returns, and offering answers to questions by using a knowledge base. It can also redirect users to a human for the rest.
- Translate knowledge base documents on the spot. The quality of machine translation is much better today than even a few years ago.
- Some service bots are tackling unstructured customer service conversations, for instance to make appointments. It sounds promising to us but it’s still early.
5. AI-enhanced CX managers
What is it?
Just like AI can help individual agents, it can also assist managers. The coaching category, above, provides examples of tasks traditionally done by managers that can be automated and extended compared to what managers can do. AI can also assist managers improve their performance and help them focus on higher-value activities.
AI can help on managers on several fronts: improving their teams’ performance with individual testing and learning paths; analyzing vast amounts of data to better forecast future performance, measure KPIs, and turn data into actionable insights; better understand skills and experience needed for success in each position, and help assess candidates for a more efficient recruiting process.
How mature is it?
In support, some manager-focused tools exist especially for escalation detection and, to a lesser extent, analyzing support activity to detect opportunities for product improvements. We believe there’s a vast opportunity here, and that the day-to-day duties of managers will change completely once the coordination function is automated.
In sales, many solutions are available and have been demonstrated for some time. They might not be cost-efficient and easy to adopt yet, but we expect that to change rapidly.
How can I use it?
Sales & Marketing
- AI-generated sales forecasts help managers with more accurate predictions.
- AI can test team members’ knowledge and skills and then help managers design personalized learning paths, identifying individual strengths and weaknesses to create a tailored training program.
- AI-generated quizzes can help sales reps and CSMs test their knowledge and identify areas of improvements when learning new topics, solutions, and industry challenges.
Support
- Case assignment for support teams that assign cases manually, as is the case in most complex support settings.
- Escalation detection for managers (we talked about this capability for support agents, above). For instance, it’s becoming possible to calculate the cost of a case in real time to take action on costly cases.
- Capacity planning, for instance forecasting staffing needs for a support team using past case history, sales forecast, and even turnover predictions.
- Developing validated requests for product changes based on customer experiences.
- Identifying high-cost customers.
- Identifying high-risk customers.
What Should You Do About These Trends?
Let’s now discuss how you can incorporate AI in your Sales, Marketing, and Support organizations this year. You may feel that your organization is not quite ready to embrace AI, or that the technical infrastructure (data, CX tech stack, IT organization) is not mature enough, along with governance. This will strongly influence how fast you can move forward or accelerate the adoption of AI for your CX teams– But you can and should act this year. Here is a framework for what actions you want to consider taking this year to take advantage of AI in your CX journey.
1. Create a strategic plan
You can’t just ignore AI completely. It’s there, it’s useful, and you will be left behind if you don’t do something. If you need help creating the plan, get help (from us, for instance), but don’t just wait for the trend to fade away. It will not!
Your AI journey will likely include a number of experiments: create a framework for where and how you will experiment.
2. Assess your foundations
Assess what AI mandate came from top management and the board. Is AI a priority? Are expectations set in terms of resources committed, and expected benefits? Is there an AI governance charter establishing principle and guardrail? What is the innovation culture in your organization and what AI research, experimentations, pilots have already taken place?
In your department (Marketing, Sales, Support, Professional Services), assess the amount and quality of the data that will be available for AI, especially to learn from customer interactions past or present. For example, are customer communications over phone, web meetings, emails, chats etc. available in text format (using to voice-to-text for phone calls and for web meetings)? Is the data structured enough to be used, or dispersed in many many siloed applications and data stores? Can you map the content of the interactions (e.g. a marketing emails, landing pages) to the customer or prospect and their attributes, and is there a success metric associated with it (e.g. successful lead conversion, or first-call resolution)?
How many, and how modern are your CRM and Marketing Automation applications? Is there any plan to retire some of them any time soon? Has your company deployed some horizontal AI productivity tool across departments (e.g. OpenAI ChatGPT, Microsoft Copilot, Google Gemini, or a custom LLM-based solution trained on your company data and business)?
How AI-aware are the different teams with your department and your organization? Do you have AI Subject Matter Experts (SMEs) within your team? If not, can you identify your future AI champions, network them together, connect with other AI champions across departments? Do you have internal or external AI innovation experts you can rely on, and will it be easy to get dedicated resources from them?
3. Don’t jump into custom tools (yet)
If you’re just starting out, don’t start by creating your own tools. It can be fun but it’s also wasteful and risky.
You can conduct your first experiment with free or affordable tools such as ChatGPT and Google NotebookLM. Then, explore the AI features of your existing CRM and communication tools. Most tools have incorporated AI-enabled features that are vastly superior to the ones available a few years ago. And you can stop here.
The next step is to consider commercial standalone tools. There is a big crop of them right now, many of them in startup mode, but you can find excellent candidates in particular for sentiment analysis and escalation detection.
If after exploring commercial tools you see that you need to create your own, then do that. But we do not recommend it for most CX organizations, with the exception of RAG LLM solutions trained on your proprietary data for deploying safe, secure tool for Enterprise Knowledge Management.
Keep in mind that AI technology is evolving very rapidly. Existing applications will add AI features rapidly in the next 12-18 months. New AI-at-the-core solutions are emerging. General AI assistants can tackle department-specific tasks (check out how Klarna claims to have moved from Salesforce and Workday to AI agents). The technology foundation for custom solutions (like RAG LLMs) is also changing rapidly. Don’t embark on an expensive 12-month project. Iterate quickly, measure success through internal learnings, enhanced customer experience, and positioning in your market as an innovator, rather than strict ROI. Don’t hesitate to revisit architecture and technical choice at every major step, but avoid paralysis!
4. Focus on a few key priorities
Most organizations should limit themselves to a handful of properties for AI. Analyze your current setup and identify the top priorities that are likely to yield the best ROI. Ignore the others.
Data is key to enable quality results. Clearly identify what data you will need, both inside and outside your organization. For instance, if you are interested in a search tool, you need a solid knowledge base, most likely your own. But if you are looking to prepare for client meetings, you will likely use plenty of external data.
5. Go step by step
Start small with a rapid Proof of Concept (PoC) on a small scale with minimal investment. Iterate as necessary before transitioning to a Pilot in a real-world environment to validate the impact of AI. Iterate again before expanding AI adoption across the organization based on learnings from the PoC and Pilot.
If you are deploying a customer-facing tool, experiment with it internally before rolling out to customers. You will want to start your alpha phase with your internal champions that are eager to adopt new technology, then roll it out to a larger audience in a dogfooding phase. Iterate until comfortable to launch a beta with selected customers. Gather insights before expanding gradually to a full launch of customer-facing capabilities. Make sure you set clear expectations for the AI tool, successfully complete AI-specific QA process (accuracy, hallucination, tone etc.), and battle-test a seamless transition to a human when AI is not able to satisfy an inquiry.
6. Manage change
We do not recommend building an AI team unless and until you absolutely know that you need to build your own tools. Conduct the experiments with your existing Ops team.
Pay attention to what IT is doing: they are likely defining AI policies and guardrails. Then, do your own thing, following policy. We like to negotiate a sandbox for any pilots so you can get them done easily and without having to ask for IT resources.
Manage change within your team. AI can appear menacing to agents and managers who fear that big chunks of their jobs may disappear (show them how the interesting part of their jobs will stay!) And remember that AI may be considered with more trepidation by older workers, who are more likely to be found in management ranks. Early positive experiences will help a lot. Be transparent about successes and defeats both.
Conclusion
AI is here, it’s working for CX, and it’s moving fast. Sit down with your team and evaluate where it could be of most use to your mission of providing a remarkable customer experience. Select a handful of priorities and investigate with small, targeted experiments, using commercial tools to start. Discard the failures. And be prepared to iterate in a few months as the landscape changes.
If any of this is too daunting, we can help!
Written by Olivier Delerm & Françoise Tourniaire
For more on AI for support, read FT’s fresh AI ideas from the ASP conference.
For more on GenAI impact across CX, read our take on the GenAI impact on the Customer Journey.