About AI agents

:::note This topic relates to using AI agents to service your customers, not how to use agents for building on StoreConnect. :::

Using the power of Einstein (Salesforce’s AI engine) and Agentforce (Salesforce’s AI Agent builder) you can add an AI agent to your StoreConnect site and increase support for your customers and users.

The benefits of using AI agents

Instead of spending time solving customer issues over the phone and using more precious human resources, an AI agent can support your customers on the frontline. If implemented well, an AI agent can:

• Reduce the demand for high touch support

• Answer common questions that might otherwise clog support channels

• Provide product information where a sales rep might not be available

• Give users the confidence to solve simple issues

• Increase profits by reducing support demands through other channels

Why use Salesforce’s AI agent tool?

You are already using Salesforce to manage your business, so why not extend your capability using a familiar platform that already knows a lot about how you work.

And because StoreConnect is built on Salesforce, we’re a gateway to using all your Salesforce information to automate and “agent-ify” your existing data.

Done right, an agent can fulfill the role of a sales rep and a support rep 24/7, in multiple languages, without you paying a cent in wages.

How do I train the AI agent?

Salesforce treats AI agents like any other Salesforce ‘user’. So it’s just like when you onboard a new user to Salesforce, and you provide them with training and permissions to access your store.

In this case, you need to provide the agent with training in the form of topics and actions, so it knows how to do a good job.

You can also set up permissions for the agent, to allow it to access and modify Salesforce data. For example, you might enable it to handle address or contact updates for your customers.

Authentication and customer identity

StoreConnect manages its own customer authentication — it does not use Salesforce as an identity provider for store visitors. This is intentional: it means your customers do not need Salesforce identity licenses or Experience Cloud licenses to use your store.

The trade-off is that when an Agentforce Embedded Service session starts on a StoreConnect site, Salesforce sees it as a guest (unauthenticated) session by default. Salesforce has well-established patterns for handling this:

• Identify known customers using hidden pre-chat fields — the simplest approach. Pass {{ current_customer.id }} as a hidden field before the session starts, then use an Omni-Channel flow to look up the Contact and associate it with the messaging session before routing to the agent. This value identifies a visitor; it does not authenticate one, so do not rely on it alone to gate agent actions that return private customer data. See Configure your store for the AI agent for step-by-step setup.

• Mid-session authentication — Salesforce’s built-in OTP flow (email or SMS) lets you verify a customer’s identity during a conversation without requiring them to log in first.

• Token-based user verification — Salesforce verifies a customer from a signed identity token. This is the only one of these options that produces a verified messaging user rather than a guest. The token is a JWT signed with RS256 or RS512, so it cannot be generated in Liquid. Produce it from a service you control, such as an Apex endpoint in your org. See Salesforce’s user verification documentation for the requirements.

:::note The HMAC-signed token described in Embed Salesforce LWC components in your storefront is a separate pattern for embedding Lightning Web Components in a storefront page. Salesforce user verification does not accept it, so it is not an option for identifying a customer to an agent. :::

Before you set up an AI agent on StoreConnect

You’ll likely need the expertise of your Salesforce Admin (if you have one) to set up an agent. But no coding skills are required. You will also need:

• A set of topics or content for feeding the agent information (we will show you how to set this up)

• A defined role and set of actions that the agent can take when interacting with customers. E.g. allow the agent to update address records in Salesforce.

• A light persona for the agent, to ensure they give customers on-brand and consistent answers

• Brand colors and elements for styling the agent

• Users ready to test the agent.