Salesforce Updates Service Cloud With Agentforce Contact Center and Casey

On October 13, 2026, Salesforce presented a series of Service Cloud updates aimed at automating customer service and bringing together the work of AI agents and contact center staff.

The main changes affected Agentforce Contact Center, the autonomous AI agent Casey, Service Rep Assistant, Workforce Engagement Management, and Customer Signals Intelligence. The company also expanded service and sales capabilities through WhatsApp and RCS and introduced new Service Cloud editions.

The updates are at different stages of the platform's development. In particular, on October 16, Salesforce opened Agentforce Contact Center in more than 30 European countries. Casey's new self-optimization capabilities were announced with a planned launch in November. The overall goal of the updates is to use CRM data to serve customers across all communication channels, automate routine requests, and give staff AI assistance in resolving complex issues.

Agentforce Contact Center: Launch of a Unified Contact Center in More Than 30 European Countries

Bringing Together Telephony, Digital Channels, CRM, and AI Agents

On October 16, 2026, Salesforce expanded the geographic availability of Agentforce Contact Center to more than 30 European countries.

The supported markets include the United Kingdom, Ireland, Germany, France, Italy, Spain, Switzerland, Portugal, Denmark, Finland, and Sweden. Previously, Salesforce's own contact center solution had been introduced in North America. The launch in Europe was the next stage of the platform's international expansion.

Agentforce Contact Center brings together:

• Inbound and outbound voice communications.
• Digital service channels.
• Routing of requests between staff and AI agents.
• Customer data and interaction history from Salesforce CRM.
• Automated execution of service operations.
• Contact center performance analytics.

The key change is the use of a single infrastructure for telephony, CRM, and AI, instead of connecting several standalone systems.

For example, a customer calls an online store and asks about the status of an order.

Agentforce can use the customer ID, purchase details, and shipping data to answer the question without an agent's involvement. If the customer needs to change the order or resolve a non-standard problem, the request is transferred to a staff member.

During the transfer, the agent receives the conversation context and CRM data, so the shopper does not need to explain the problem again. For companies with a large volume of requests, this can reduce handling time and lower the workload on agents.

International Deployment That Takes Local Data Requirements Into Account

In the October announcement, Salesforce specifically emphasized requirements for data residency, regulation, and information security. For international organizations, this is especially important, since contact centers process customers' personal data, call recordings, and information about commercial transactions.

Agentforce Contact Center is designed to operate in several countries while keeping a shared CRM infrastructure.

For example, an international company can organize customer support in France, Germany, and Spain through a single Salesforce system while keeping regional settings for communication channels.

At the same time, the platform's availability in a particular market does not mean it automatically meets all the requirements of a specific organization.

Before implementation, you need to separately check the terms for storing recordings, processing personal data, using phone numbers, and transferring information across borders. In November 2026, Salesforce planned to expand the availability of Agentforce Contact Center to Australia and New Zealand.

Setting Up a Phone Contact Center Through AI Tools Without Manual Configuration

In October, Salesforce also introduced new Headless Skills for Agentforce Contact Center.

They are designed for setting up phone infrastructure through AI development tools, including Claude and Cursor. Instead of performing each action manually in the Salesforce interface, an administrator can use a specialized Skill.

For example, the task may include:

1. Selecting the country and phone number type.
2. Choosing an available number.
3. Connecting the number to Salesforce.
4. Creating a voice channel.
5. Setting up routing.
6. Connecting an AI agent.
7. Creating a fallback route for transferring the call to a staff member.

Two main routing options are provided.

Omni Queue sends incoming calls directly to the contact center queue. Omni Flow makes it possible to use an AI agent and a defined routing scenario, including a subsequent transfer to an agent.

Thus, the administrator can automate part of the technical setup of the contact center.

However, using Headless Skills does not remove the need to verify permissions, the phone infrastructure, and the correctness of the routes created.

At the time of the October update, the corresponding tool was in beta.

Casey: An AI Agent for Resolving Customer Requests on Its Own

Handling Returns, Account Questions, and Routine Service Requests

In the October overview, Salesforce separately presented the development of Casey, an autonomous AI agent for customer service. Casey is part of the portfolio of specialized Agentforce Agents that the company introduced on September 11, 2026.

Its purpose is to handle shopper requests on its own across several communication channels.

Casey supports:

• Answers to frequently asked questions.
• Checking order and account information.
• Handling defined return scenarios.
• Performing permitted service operations.
• Transferring complex requests to staff.

The AI agent can interact with customers through voice channels, SMS, WhatsApp, and web chat.

For example, a shopper contacts the store about returning a product.

Casey identifies the customer, retrieves the purchase details, checks the applicable return rules, and offers a permitted way to resolve the issue. If the return complies with the store's rules, the agent can perform the defined actions.

If the situation requires an individual decision, the agent transfers the request to a staff member. Unlike a regular chatbot, Casey is designed not only to provide information but also to perform actions using CRM data and configured business processes.

At the same time, the agent's capabilities are determined by the permissions granted to it.

Casey's Self-Optimization Based on Real Requests

On October 13, Salesforce announced the next stage of Casey's development, the Self-Optimizing AI Agent. The company planned the launch of the new capabilities for November 2026. The main change is the automation of the process of improving the AI agent itself.

In the traditional model, the Customer Service team itself analyzes requests, identifies recurring problems, creates new automation scenarios, and tests them before implementation. The new Casey system is meant to perform a significant part of this work.

The process includes several stages.

1. Request analysis. The system examines real conversations and service cases handled by AI agents and staff.
2. Identifying automation opportunities. Casey groups requests by reason and assesses which of them are best suited for the AI to handle on its own.
3. Prioritization. Automation opportunities are ranked based on request frequency and the readiness of processes for automation.
4. Creating and improving scenarios. The agent uses existing data, rules, and Salesforce Flows to prepare new capabilities.
5. Testing. Changes are tested on scenarios that reflect real customer requests.
6. Approval. Each change requires a staff member's confirmation before it is published.

For example, a service company finds a large number of requests about rescheduling service appointments. Casey can identify the recurring scenario, prepare automated handling, and test it on historical examples. After review, a staff member can approve the new scenario.

Important: self-optimization does not mean that the AI gets the right to change business processes on its own without oversight. Salesforce requires mandatory human approval of changes.

As of November 8, 2026, the company had announced the release of the new functionality during November but had not specified an exact date for general availability.

Transferring a Request to a Staff Member While Keeping the Context

An important part of Casey's work is interaction with agents. If the AI cannot complete the service on its own, the request is transferred to a staff member. Along with the request, the interaction history, a conversation summary, and a recommended next action can be passed on.

For example, if a customer asks for compensation that exceeds the permitted limit, Casey must transfer the decision to an authorized staff member.

This mechanism makes it possible to separate standard operations from situations that require a human decision. For businesses, this is especially important for returns, financial transactions, changes to service terms, and handling complaints.

Service Rep Assistant: An AI Assistant Builds a Resolution Plan During the Conversation

Dynamic Instructions for Staff Based on the Context of the Request

On October 13, Salesforce presented updated capabilities of Service Rep Assistant, an AI assistant that works directly in the service interface.

Unlike Casey, which can communicate with the customer on its own, Service Rep Assistant is designed to support the staff member. It analyzes the request data, the interaction history, CRM information, and the available materials in the corporate knowledge base.

Based on this information, the system builds a step-by-step plan for resolving the problem.

The main change is Dynamic Plans, which are updated as the conversation develops.

For example, a customer contacts a SaaS company's technical support with a complaint that they cannot connect an integration.

Service Rep Assistant examines the account information and suggests a diagnostic sequence. During the conversation, it turns out that the problem occurs only for users with a certain access role.

The AI updates its recommendations to take the new circumstance into account. The agent receives a refined plan without starting the diagnosis over. Unlike static instructions, Dynamic Plans adapt to incoming information.

Support for Case and Messaging Sessions

The new Service Assistant works both with request records in Salesforce and with active messaging sessions. In Case mode, the AI tracks changes to the record, including new comments and field updates.

In Messaging mode, it analyzes the current conversation and adapts its recommendations in real time.

Depending on the configuration, two types of plans are used.

Guidance Plans: pre-built instructions that the staff member can update manually. Dynamic Plans: dynamic recommendations that change as new data comes in.

Messaging Sessions work with Dynamic Plans. In the Service Assistant interface, the staff member can view suggested actions, ask the AI for additional information, and use the available Agent Actions.

At the same time, the AI assistant must not automatically perform operations that go beyond the staff member's permissions.

Agentic Slack Swarming: Bringing In Experts Without Leaving the Service Console

On October 13, Salesforce also announced the expansion of Agentic Slack Swarming.

The feature is designed for complex requests that require the involvement of several specialists.

For example, a customer of a B2B SaaS company reports an integration problem that affects the API, billing, and access management at the same time.

The agent needs to bring in a developer, a technical support specialist, and a member of the finance team. Service Rep Assistant can start a separate Slack channel, add the necessary participants, and pass on the context of the request.

The discussion uses Case data and the available AI tools. This allows staff to solve the problem together without losing information between different systems.

Feature status: in October, Agentic Slack Swarming was in open beta.

To connect it, you need to link the Salesforce Organization with a Slack workspace and configure permissions. Salesforce Setup includes a Swarming section where you can activate Use the New Swarming in Slack (Beta).

You also need to set up Salesforce Channels and grant the corresponding permissions to staff.

WhatsApp and RCS: Expanding Sales and Support Inside Messages

WhatsApp: Completing Purchases and Payments Without Leaving the Conversation

In the October announcement, Salesforce expanded the capabilities of the WhatsApp integration with Service Cloud. There are now scenarios in which the customer can not only get a consultation but also make a purchase directly in the conversation.

For example, a shopper contacts a store through WhatsApp and asks whether a product is in stock.

A staff member or AI agent can help with the choice, clarify the specifications, and guide the customer through the supported ordering process. Depending on the integration configuration, scenarios for processing payments inside the conversation are available.

This expands the role of WhatsApp as a channel for Customer Service and Conversational Commerce. For E-commerce companies, this functionality is especially relevant when selling products that require a consultation beforehand.

However, connecting WhatsApp to Salesforce does not mean that all payment functions are automatically available.

To use commerce scenarios, you need to check support for the corresponding message types, the payment infrastructure, and regional restrictions.

RCS Messaging: Product Cards, Interactive Messages, and Two-Way Communications

In October 2026, Salesforce announced the general availability of Rich Communication Services (RCS) in supported markets.

RCS expands the capabilities of traditional SMS and allows the use of interactive elements. In particular, high-quality images, cards, carousels, buttons, and suggested replies are supported.

For marketing, this opens up additional options for preparing messages about products, promotions, and customer events.

For example, an online store can send a shopper a message with several recommended products.

Unlike a regular SMS, RCS makes it possible to present products as interactive cards. In services, RCS can be used to confirm appointments, reschedule service times, and send information about upcoming visits.

In two-way scenarios, customer replies can be processed through Agentforce.

On October 28, Salesforce published additional technical recommendations on RCS.

The company stated that the functionality is available in, among other countries, the U.S., the United Kingdom, Germany, France, Mexico, Brazil, and India. Availability depends on the country, carriers, supported devices, and the selected message type.

Workforce Engagement Management: Joint Planning of Work for Agents and AI Agents

A Single System for Managing Human and Automated Resources

On October 13, Salesforce presented additional capabilities of Workforce Engagement Management (WEM) for managing contact centers.

Traditional Workforce Management systems calculate staff workload, build work schedules, and monitor schedule adherence. As AI agents spread, some requests begin to be handled automatically.

As a result, a manager needs to take into account not only agents but also the volume of work performed by AI. The updated WEM brings this data together in a shared management system.

Contact center managers can monitor staff workload, schedule adherence, and team performance metrics. In addition, Salesforce offers AI recommendations for work planning.

For example, a manager can request a workload forecast for the next week and get recommendations on staff allocation.

Forecasting Staffing Needs With Salesforce Moirai

For forecasting, Workforce Engagement Management uses Salesforce Moirai technologies. They are designed to analyze time series and determine future workload based on historical data.

For example, a service company can estimate the expected number of calls and messages on different days of the week.

The calculations need to take into account not only the total volume of requests but also the share of requests that can be handled by AI. This makes it possible to plan the required number of staff based on the actual workload.

Salesforce also supports managing operations through natural language queries.

For example, a manager can ask the system to estimate staffing needs for a certain period.

However, forecast results should be checked, since actual workload depends on seasonality, marketing campaigns, technical incidents, and other factors.

WEM can be used both with Salesforce's own Agentforce Contact Center and with supported partner contact centers.

Customer Signals Intelligence: Analyzing Sentiment and Reasons for Requests

Assessing Customer Sentiment Directly in Command Center

In October, Salesforce expanded the capabilities of Customer Signals Intelligence.

The new features make it possible to analyze the sentiment of customer requests and receive the corresponding signals inside Command Center. Instead of examining each conversation separately, the manager receives aggregated information about the nature of the requests.

The system uses data from Cases, voice conversations, and digital communications.

For example, after launching a new subscription model, a SaaS company may find an increase in negative requests about payment.

Customer Signals Intelligence helps detect changes and determine which categories of problems need attention. In Salesforce's official release notes, the corresponding updates were added in the week of October 19, 2026.

Topic Clustering: Automatically Identifying the Main Reasons for Requests

Another change is the automatic grouping of requests by topic. Topic Clustering makes it possible to classify conversations and identify the most common reasons for contacting support.

For example, for an E-commerce company, these categories might be:

• Shipping delays.
• Product returns.
• Payment errors.
• Product out of stock.
• Problems applying a discount.

The system ranks topics by request volume. This allows the manager to determine which problems create the greatest workload for the contact center. The resulting data can be used to develop the knowledge base, improve service processes, and prepare new Casey scenarios.

For example, if a significant share of requests relates to checking order status, the company can implement a corresponding automated scenario.

At the same time, automatic classification must be checked, especially when handling ambiguous requests and messages in different languages.

New Core, Advanced, and Max Editions: Bringing Together CRM, Contact Center, and AI

Ready-Made Product Bundles Instead of Separate Licenses

On October 13, Salesforce announced new Service Cloud editions: Core, Advanced, and Max.

The goal of the change is to simplify purchasing and using products that the company previously could license separately. The new model includes options for combining Service Cloud, the contact center, Workforce Engagement Management, Slack, Tableau Next, security tools, and support.

The specific set of features depends on the selected edition. This is important for large companies that use Salesforce at the same time as a CRM, a service platform, an analytics system, and an environment for AI agents.

Instead of a large number of separate products, an organization can choose a more comprehensive offering.

Flex Credits: Included Resources for Using AI

Salesforce included Flex Credits in the new editions, which are used to pay for defined Agentforce operations.

The announced amounts:

These credits are intended for using AI functionality within the selected offering.

However, the number of credits does not correspond to a fixed number of conversations with customers.

Actual consumption depends on the type of operations performed and Salesforce's current pricing model. Therefore, when choosing an edition, you need to estimate the expected volume of requests, the share of automated service, and the cost of using AI.

In the October announcement, Salesforce did not present a universal price that would make it possible to compare the full cost of all three editions without an individual calculation.

How to Prepare Salesforce Service Cloud to Use the Updates

To implement the new capabilities, you need to start by checking the current Salesforce configuration.

Stage one: CRM and customer data. You need to make sure that Case, Account, Contact, and the other objects used contain up-to-date information.

If data on orders, contracts, or requests is incomplete, AI agents will not be able to handle the corresponding requests correctly.

Stage two: the knowledge base. You should check the instructions for returns, payment, technical support, and other service operations.

Casey and Service Rep Assistant use existing business data and rules, so the quality of the knowledge base directly affects the quality of the recommendations.

Stage three: communication channels. You need to determine which channels the company plans to use: telephony, WhatsApp, RCS, web chat, or other supported integrations.

Stage four: AI permissions. You should separate the operations that Casey can perform on its own from the actions that require a staff member's confirmation.

For example, providing information about order status can be automated, while a refund of a large amount must go through an approved procedure.

Stage five: testing. Before launch, you need to test routine requests, non-standard situations, and the transfer of conversations to staff.

For this, Salesforce provides Agentforce Testing Center.

Stage six: measuring performance. After launch, you should assess the share of automatically resolved requests, repeat contacts, handling time, customer satisfaction, and support costs.

At the same time, a decrease in the number of requests to agents should not automatically be considered an improvement in service quality. You need to check whether customers' problems have actually been resolved.

Practical Value for E-commerce

For online stores, the most important aspects are automating order-related requests, bringing together messages from different channels, and supporting commerce transactions inside conversations. Casey can handle routine questions about shipping, returns, and product specifications.

Service Rep Assistant helps staff solve non-standard order problems. WhatsApp and RCS expand the options for communicating with shoppers. Customer Signals Intelligence makes it possible to identify recurring reasons for customer dissatisfaction.

For example, after launching a large advertising campaign, an online store may see a significant increase in requests.

AI agents make it possible to handle part of the standard requests without a proportional increase in the number of agents.

At the same time, full automation requires integrating Salesforce with the commerce platform and the order management system.

Practical Value for Customer Services

For service companies, the main value lies in workload management and automating recurring requests.

For example, a company that provides financial, educational, or professional services can use Casey to handle standard questions.

Staff continue to handle complex customer situations. Service Rep Assistant helps them carry out defined procedures, and Workforce Engagement Management makes it possible to plan the required number of agents.

In addition, Customer Signals Intelligence helps identify the reasons for growth in requests and monitor service quality.

Practical Value for B2B SaaS

For B2B SaaS companies, the most important elements are Service Rep Assistant, Slack Swarming, and the use of CRM data in technical support.

For example, a customer reports a problem with a product integration.

Service Rep Assistant uses account data and request history to prepare a diagnostic plan. If the problem requires developers' involvement, the agent can start collaboration through Slack.

The information about the problem is kept in Salesforce. Casey can be used to handle standard questions about settings, access, and product functionality. This approach helps bring together Customer Support, Customer Success, and technical teams.

Availability and Limitations

The changes announced by Salesforce have different statuses. Agentforce Contact Center became generally available in more than 30 European countries on October 16, 2026. The availability of specific phone capabilities needs to be checked by country and connection terms.

Casey was introduced earlier, on September 11, 2026. The new self-optimization capabilities have been announced for November. Service Rep Assistant is being developed as part of Service Cloud and provides dynamic recommendations to staff in supported configurations.

Agentic Slack Swarming was in open beta. RCS became generally available in October in certain markets.

Workforce Engagement Management and Customer Signals Intelligence expand the capabilities for contact center management and request analytics. Core, Advanced, and Max are new options for bundling Salesforce products, and their availability and composition need to be checked when signing a contract.

Salesforce Service Cloud: What the Updates Mean

Salesforce Service Cloud is a customer service management platform that brings together CRM, request handling, communication channels, automation, and analytics. The October updates expand its capabilities in three main areas.

The first is bringing together the contact center and CRM, so that requests are handled using a shared customer context. The second is automating service through Casey and supporting agents with Service Rep Assistant.

The third is managing the joint work of staff and AI agents through Workforce Engagement Management and Customer Signals Intelligence. The updates are aimed primarily at mid-sized and large companies with a high volume of requests, international divisions, and complex service processes.

For such organizations, the practical value lies in automating recurring operations, reducing request handling time, and allocating contact center resources more efficiently.

Sources

1. 1. Salesforce, October 13, 2026: What's New in Service Cloud: 10 Launches That Help Resolve Customer Service Cases Faster
2. 2. Salesforce, September 11, 2026: Salesforce Expands Agentforce With a New Portfolio of AI Agents
3. 3. Salesforce: Agentforce Contact Center
4. 4. Salesforce Help: Agentforce Contact Center: Customer FAQ
5. 5. Salesforce AgentExchange: Agentforce Contact Center Coordinate Skill
6. 6. Salesforce Trailhead: Explore the New Service Assistant Dynamic Experience
7. 7. Salesforce Help: Agentic Swarming with Slack in Agentforce Service
8. 8. Salesforce Help: Agentforce Testing Center
9. 9. Salesforce Help: Release Note Changes, October 2026
10. 10. Salesforce, October 28, 2026: RCS Messaging on Salesforce: A Practitioner's Guide
11. 11. Salesforce: Agentforce Contact Center Pricing
12. 12. Salesforce Help: Offer Native Messaging Support with Agentforce Contact Center