How to Prepare Your Service Team for Digital Labour: A Practical Guide

The maturity of your data management will affect your Agentforce implementation. The more comprehensive and well-structured your data, the better your AI agents will perform.

But whether you have a mature data strategy or your organisation is at the beginning of its data journey, Agentforce still has the power to transform your service experience. Begin by focusing on a specific area or department where the impact of AI can be most immediately felt. This could be a high-volume customer service line or a frequently asked questions (FAQs) section. By starting small, you can test the waters, gather valuable insights and make the necessary adjustments before scaling up.

The adage “rubbish in, rubbish out” holds true for AI. If your data is messy, incomplete or inconsistent, the responses from your AI agents will be equally problematic. Take the time to clean and organise your data. This includes removing duplicates, correcting errors and ensuring all information is up to date and accurate. The good news: You don’t have to clean up all your data to start. Instead select a small subset to focus on while you work on a larger clean-up strategy.

Bringing all your data into one unified platform is crucial for creating a cohesive and comprehensive knowledge base. This means integrating data from various sources, such as:

• Customer service logs
• Product manuals
• User feedback
• Customer satisfaction surveys
• Support tickets
• Purchase history

A unified platform ensures your AI agents have access to the most relevant and up to date information, leading to more accurate and helpful responses. For example, Unified Knowledge makes it easy to bring all of your third-party data directly into Salesforce to help make AI grounding more effective.

Your knowledge base holds the key to making AI-generated content more accurate and specific to your organisation. This is done through a process called AI grounding. Grounding (commonly done by retrieval-augmented generation (RAG)), ensures that your AI-generated content is both accurate and unique.

To get the best results from AI grounding, you need a well-organised and comprehensive knowledge base. When creating and maintaining your knowledge base, make sure you provide detailed, complete information. AI works best when it has enough context to generate varied responses, depending on the situation. Using real-world examples can also help AI understand how to apply knowledge in specific situations — whether that’s troubleshooting an error or resolving a common customer issue. And separating customer-facing content from internal procedures ensures that AI delivers the right information to the right audience.

Also, don’t forget about media. Adding descriptive text and alt tags to images or videos helps AI interpret visual content and makes your knowledge base more accessible to users with disabilities. Writing FAQs that directly address common customer questions is another simple but effective way to improve AI’s performance. And when AI can easily match a question to a well-written answer, it becomes much more efficient in delivering helpful responses. Start with a handful of common questions and as your knowledge base grows, so, too, will AI’s ability to assist with more complex cases.

To keep your Agentforce agent learning and adapting, you’ll need to regularly monitor your data and identify knowledge gaps. Address these gaps to ensure your AI agents are always providing the most accurate and relevant information. You can use these five methods to make sure your knowledge is up to date:

1. Regular audits: Conduct monthly reviews of your data sources to ensure that they are up to date and comprehensive. This can help to identify missing or outdated information.
2. Customer feedback: Analyse customer feedback from surveys, reviews and support interactions to pinpoint areas where customers feel the information is lacking or unclear.
3. Employee feedback: Gather input from your customer service team. They often have firsthand experience with customer questions and can highlight areas where more information is needed.
4. Frequent questions and support tickets: Review the most common questions and issues that come up in support tickets. If certain questions are repeatedly asked, it may indicate a need for better documentation or training.
5. Cross-department collaboration: Work with product development, engineering and marketing to ensure all relevant information is being captured and shared.

It’s not enough to just fill gaps. Also make sure your AI agents are regularly updated with the latest information and are continuously learning from new data. For example, if your company starts a new promotion where selected products are 50% off, your AI agents can use this information to upsell and drive new business for the company.

To get started, try making sure all new information and data hits your Agentforce agents on a weekly basis. And add this data integration into any important launch plan, surge strategy or changes to the business. This will help your Agentforce agents stay in line with your business’ priorities and enhance the overall customer experience and the efficiency of your service operations. That way, your service organisation can help drive more revenue for your business — in addition to providing better customer service.

Use cases and testing are important for AI agents to ensure that they perform effectively, reliably and meet specific customer needs. But you don’t need to pilot everything at once. Start by identifying the top reasons why your customers are contacting you. Start small — review your organisation’s case history from the past six months to identify your customers’ top 10 contact reasons. Later, you’ll be building AI agents to effectively handle these 10 contact reasons. This data-driven approach will help you to focus on the most common and critical issues your customers face. By addressing these high-frequency contact reasons, you can ensure that your AI agent is immediately useful and provides value from the outset.

Once you've identified your organisation's top 10 contact reasons, create a detailed knowledge article for each one. These articles should cover the most common questions, issues and solutions related to each contact reason. You can even save time by having generative AI create the articles based on analysing previous case details. (Just make sure that you have a human review for accuracy.) When creating these articles, focus the content on a singular topic vs. multiple and make it structured. This makes it easier for AI to review and process for an accurate answer. Each article should include:

• A clear title: Ensure the headline is clear, concise and accurately describes the content within the article. This also makes the article easier to find.
• Bulleted lists or subheadings: Make your knowledge article easy to skim. Create bulleted lists or subheadings that show what’s covered in each section. If you’re writing a tutorial, write step-by-step instructions — including which buttons or menu items to click — and put them in chronological order.
• Images, links or videos: Give as much detail as you can so both Agentforce and service reps can successfully solve the issue on their own. Add screenshots, images, links to other articles or videos to your article to help them.
• Simple language: Write in a way that can be easily understood by everyone. Do your best to avoid jargon that specifically relates to your product and service. Your agents won’t have as much product knowledge as you, so use simple language in each step of your article.

Use these knowledge articles to define the rules, actions and intentions for your AI agent. If you’re not able to build a knowledge article, the AI agent should be programmed to guide the user to the most relevant information or escalate the issue to a human rep if necessary.

Consider releasing your Agentforce agent to a small group of users initially. This pilot phase allows you to gather feedback and make adjustments before a full-scale roll-out. Make sure you solicit immediate feedback from customers (and reps) immediately after their Agentforce experience and action on any trends. This approach helps you to identify and address any issues early on, ensuring a smoother and more successful implementation.

Testing is crucial to ensure your Agentforce agent can handle a wide range of scenarios. Develop a comprehensive testing plan that includes both common and edge cases. Use real-world scenarios to simulate different customer interactions and evaluate the AI agent’s performance. Pay attention to how the AI handles complex queries, emotional responses and unexpected inputs. Regularly update your testing scenarios to reflect new data and evolving customer needs — perhaps quarterly. By thoroughly testing your AI agent, you can build a robust and reliable system that meets the needs of your customers and service reps.

When considering where to deploy your AI agents, more channels doesn’t always equate to a better customer service experience. Let’s dive into how you decide where to implement Agentforce.

Your Agentforce channel strategy should align with your business goals and KPIs. For example, if you’re a retailer with high volumes of low-complexity interactions, your goal might be to reduce the need for large support teams and lower operating costs. In this case, investing in digital channels (such as chat and messaging) and automation with AI agents would be a strategic move. On the other hand, if you are in an industry like financial services or healthcare, where customers expect a higher level of service and deal with complex issues, the channels you choose and the level of automation you invest in will look quite different. You might want to prioritise human-assisted support for high-value interactions over deflecting those cases.

Understanding your contact drivers — why customers keep in touch to your organisation — is crucial. Map out the customer journey for each contact driver to identify the most efficient and effective channels for resolution. For instance, for common issues such as a simple password reset, a self-service portal or an AI agent might be the best fit. For more complex issues, such as financial advice or medical consultations, a combination of AI agents and human support might be necessary.

Determine which channels are the best fit for each contact driver based on the level of support needed and the cost to serve. For low-complexity issues, prioritising self-service portals and digital channels are cost-effective and convenient. For high-complexity issues, a hybrid approach that combines AI agents with human support can provide the necessary level of service while still maintaining efficiency. By carefully selecting and optimising your channels, you can ensure that customers are directed to the most appropriate and efficient resolution path, enhancing their overall experience and satisfaction.

The key is to guide customers to the channel that is best for their specific needs. The best channel is one that is the most efficient and requires the least amount of customer effort, while still resolving their issues effectively. Speed and convenience are paramount in customer service, especially in digital channels. Customers value quick and seamless interactions and your channel strategy should prioritise these factors.

When rolling out Agentforce, it's essential to communicate to your customer service teams how AI will augment their capabilities rather than replace them. Emphasise that AI will help them to get more done by handling routine and low-complexity tasks, allowing them to focus on more strategic and high-value interactions. Their human skills, such as empathy, problem-solving and relationship-building, are still very much needed to provide a great experience for your customers. By using AI, your human teams can do more human things; they can operate more efficiently and effectively, leading to higher job satisfaction, better customer outcomes and less burnout.

Provide regular, timely opportunities for feedback from your human teams. Encourage them to share their experiences and insights on how the AI agents are performing and where improvements can be made. Use this feedback to iterate on the guardrails for your AI agents, ensuring that they are aligned with your business goals and customer needs. Continuous improvement is crucial to maintaining the effectiveness and reliability of your AI system.

Agentforce disrupts customer service by seamlessly integrating powerful insights from Customer 360, merging sales, marketing and commerce data into one unified view. With Agentforce’s ability to analyse and present this data (using Data Cloud), agents can close cases faster, make smarter decisions and offer personalised solutions that truly resonate with customers. The result? More opportunities for upselling and cross-selling as service reps can make highly relevant recommendations based on a deep understanding of each customer’s needs and history. AI is not just a tool for automation; it’s a powerful ally that both empowers service reps and enhances the customer experience.

Ensure that the handoff between AI and human agents is not only transparent to the customer, but also seamless in terms of both speed and data transfer. When an Agentforce agent needs to escalate an issue to a human rep, the transition should be swift, with all relevant information — from prior conversations to customer history — immediately available to the rep. This ensures customers don’t have to repeat themselves or reexplain their issue, reducing frustration and improving satisfaction. For the human rep, having the complete context at their fingertips means they can hit the ground running, resolving the case without wasting time getting up to speed. This smooth, data-driven handoff not only fosters customer confidence but also accelerates case resolution, ensuring that both the customer and the agent are aligned and focused on the same goal: a fast, efficient solution.

Back to the top

When integrating Agentforce into a customer service operation, performance metrics become essential for tracking its success and understanding areas of improvement. While there isn’t a one-size-fits-all approach, there are some key metrics you can monitor to gauge the effectiveness of this new technology.

• Case resolution: A good starting point is the automated resolution rate, which reflects how well Agentforce handles customer enquiries without needing human intervention. Before deploying Agentforce, it’s critical to document the current resolution rate so there’s a baseline for measuring improvements. The goal is to improve this rate by at least 30%, showcasing the impact of the new system.
  
  For instance, Salesforce’s Help page boasts over 740,000 articles, but its breadth can make it challenging to find the right answer quickly. Agentforce addresses this by guiding users to the most relevant answers, solving 83% of cases without requiring a human agent.
• Cost savings: Agentforce automates routine customer interactions, reducing the need for large human teams. Additionally, it reduces training expenses, as the digital labour platform continuously learns and handles common queries, cutting down on onboarding costs. By optimising resource allocation and providing data-driven insights, businesses can further improve efficiency, ensuring support is scaled effectively without unnecessary overhead. These combined efficiencies help organisations streamline operations, reduce costs and enhance profitability. Agentforce customers are projected to reduce cost of service by 25% or more by year three.
• Human resource reallocation: Show how reallocating service reps can improve other aspects of your business. Perhaps your reps are currently spending a lot of time on low-value tasks - like resetting account passwords. Once that’s automated with Agentforce, your people resources can be reallocated to more complex cases or spend more time on revenue-generating opportunities.