Customer service is moving so fast, and with the help of AI, it will eventually play a massive part in future interactions. Agentforce by Salesforce is an uber-powered tool that helps businesses make intelligent AI service agents that automate their support workflows and provide excellent customer experiences. This article explores in depth the features of Agentforce with how it connects with Salesforce Data Cloud and its impact across Salesforce Service Cloud, Sales Cloud, and Marketing Cloud from an instance-tested proof point recently where they built agents that could deflect cases and build technologies such as scheduling installations.
Agentforce: Beyond Traditional Chatbots, Powered by Data Cloud & Integrated Across Clouds
- Agentforce goes beyond the boundaries of conventional chatbots by defining agent behaviors and responsibilities using natural language processing. Agent responsibilities differ from traditional chatbots. Its power is enhanced by the seamless weaving throughout Salesforce Data Cloud. When placed within the greater Salesforce eco-system, its influence is greatly improved as it speaks to the Service Cloud, Sales Cloud, and Marketing Cloud of the rest.
- Agentforce, when Data Cloud and other clouds are wired, allows users to build tasks & guardrails for agents with human language using the same customer data platform that informs AI decisions, removing silo barriers. Unlike in the old systems rigid dialog trees that connected our agents, Agentforce is connected to Data Cloud.
Key Features and Functionalities:
Topic Definition:
- Topics are the building blocks; they set what an AI agent can and cannot do.
- Users can set up what jobs an agent will do and won’t, defining “guardrails” for how it should act.
- Instead of hard-to-read rules and logic, natural language descriptions derived from the deep data of the Data Cloud (which brings together service cloud, sales cloud, and marketing) make it easier to define agents’ tasks.
Action Creation:
- Actions allow the agents to take action through Salesforce Flows, Apex classes, or prompts.
- Flows can be utilized for information retrieval (e.g., fetching available install times from Service Cloud) and to perform an action (e.g., setting the lead status in Sales Cloud, launching Marketing Cloud journeys), updating data in the relevant worlds of Data Cloud and Cloud.
- This enables pre-built processes from Salesforce to be integrated into agent workflows, with the data cloud acting as a single source of truth across all Salesforce clouds.
Contextual Understanding:
- Agentforce employs a conversational context to provide more understandable responses to complex user requests, using the rich context available through the Data Cloud (all interactions from the Service Cloud, purchase history from the Sales Cloud, and marketing engagements in the Marketing Cloud).
- It can understand vague statements and retort appropriately, showing very deep contextual comprehension.
- Agentforce plans and performs the actions dynamically throughout the conversation, using the unified customer profile in Data Cloud as an execution context.
Planning and Execution Dynamically:
- Agentforce generates dynamic plans in real time according to the context of the data from the Data Cloud.
- Execute flows, data shaping, and response filtering by conversation to front-end the workflows of all clouds (Service Cloud, Sales Cloud, and Marketing Cloud), ensuring accuracy and relevance.
Guardrail Enforcement:
- Agentforce provides guardrails (pre-defined rules) to stop agents from doing things beyond their authority.
- Evaluate off-topic requests and take users to a pre-approved topics list-controlled interaction or off the site.
- It stops actions outside defined ranges, e.g., data validation based on Data Cloud (which can be sourced from Sales Cloud delivery dates competing with our application), until an installation is scheduled after delivery.
Security and Safety:
- Built-in harm and toxicity detection to keep the agents from engaging in irrelevant or malicious conversations with customers.
- Detects and rejects prompt injection attacks, faces declining recrudescent attempts to reverse engineer.
Multi-Channel Deployment:
Agentforce (powered by Data Cloud’s unified data) allows the distribution of AI agents to many channels for full-blown agent conversation in Service Cloud, Sales Cloud, and Marketing Cloud interactions on an end-to-end consistent customer experience.
Building an Installation Scheduling Agent with Data Cloud and Cross-Cloud Integration:
This demonstration showed how Agentforce created an agent scheduling the installation appointment. New “appointment management” topic (with associated actions- Salesforce flows) on the agent side, so they could
- Get free slots for installation, pulling real-time scheduling data from the Service Cloud through the Data Cloud.
- Automate bookings (updating customer records and scheduling systems in the data cloud and, who knows, maybe sending some follow-up campaigns to the marketing cloud as well)
- Validate user input (Installation dates must be after delivery dates) by consuming data from the Sales Cloud in the Data Cloud through Customer Orders and Deliveries.
- Resolve vague requests (e.g., “That Friday, okay? but Biffy Morning?”) by utilizing past service requests, sales history from Data Cloud (past campaigns engagement) stored in the customer preferences, and any prior interactions.
The Power of Natural Language and Unified Data Across Clouds:
The natural language descriptions used by Agentforce and leveraging Salesforce Data Cloud + integration capabilities in Service Cloud, Sales Cloud, and Marketing Cloud are key to maintaining developer simplicity dialog treeless experience. Most importantly, allowing an agent to speak with customers is a deep well of insights. This enables users to author nuanced agent behaviors based on plain language with an immovable customer table to conveniently deliver the proper replies at scale in all customer touchpoints.
Conclusion:
Salesforce Agentforce, when connected with Salesforce Data Cloud and all these are synced up in Service Cloud, Sales Cloud, and Marketing Cloud, is a step ahead in terms of customer experience powered by AI. Businesses that leverage natural language processing, strong guard rails, and a CDP unified customer data can now build at-scale intelligent agents that automate processes and improve customer experiences at all stages of the selling journey and company-wide for any customer-facing department. This harmony will only heighten as AI matures and the marrying of intelligent automation with unified data is achieved in customer relationship management.
