> For the complete documentation index, see [llms.txt](https://docs.eesel.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.eesel.ai/integrations/helpdesks/salesforce.md). # Salesforce Set up an AI agent for Salesforce Service Cloud. It reads cases, accounts and Knowledge, drafts replies as case comments, and keeps your records up to date. eesel's AI agent for Salesforce Service Cloud works inside your Salesforce org. It reads cases and Knowledge articles (and contacts and accounts once you switch them on), drafts replies as internal case comments, and updates records. Salesforce holds the customer record. Customers rarely write in through Salesforce, so eesel mostly uses it to look up context: which plan the account is on, what is still open, and what your team already told them. Your AI teammate reads cases and knowledge articles, plus contacts and accounts once you switch them on (see [Sources](#sources)), plus anything else you connect, such as your help center, Confluence, Notion or your website, and works with the record in front of it. It can create and update records too, so a conversation ends with Salesforce in the right state instead of a note for someone else to type up later. In this example the agent read the case, looked up the account, searched twelve other cases, and posted a private case comment.
A Salesforce case about scanners failing after a firmware update, with eesel AI posting a private case comment after reading the case, looking up the account and searching twelve other cases

Your agent works the case with the record in front of it, looking up the account and past cases before it says anything.

{% hint style="info" %} **Connecting Salesforce has one extra step.** eesel's package has to be installed in your Salesforce org before you authorize the connection. Your agent gives you the install link, then the authorization link. The steps below walk through both. {% endhint %} ## Set up eesel in Salesforce {% stepper %} {% step %} ### Ask your agent to connect Salesforce Open the chat in your dashboard and ask it to connect Salesforce. It replies with two steps. First install the eesel AI package in your Salesforce org, because Salesforce needs the package before it allows the connection. The agent gives you an **Install eesel AI package** link. Prefer to click? **Integrations** in the left sidebar has the same option.
The dashboard chat where the user asks to connect Salesforce and the agent explains the package must be installed first, with an Install eesel AI package link

The agent posts the package install link in the chat.

{% endstep %} {% step %} ### Install the eesel AI package in Salesforce Open the link as a Salesforce admin. If you are not signed in, Salesforce asks you to sign in first.
The Salesforce login page, asking the user to log in before the install page opens

Salesforce asks you to sign in before it opens the install page.

The install page then opens. Choose who gets the package: admins only, all users, or specific profiles, whichever fits your team. Tick the box to acknowledge the app, then click **Install**.
The Salesforce Install eesel AI page, with Install for Admins Only selected, the acknowledgement box ticked, and the Install button

The Salesforce install page for the eesel AI package.

Salesforce then asks you to approve the site the package talks to. That site is `oracle.eesel.app`, which is eesel. Tick **Yes, grant access to these third-party web sites** and click **Continue**.
The Salesforce Approve Third-Party Access dialog listing oracle.eesel.app, with the grant access box ticked and a Continue button

Salesforce asks you to allow the package to reach eesel.

Salesforce confirms when the install is done. The page shows the package name eesel AI and its version. Click **Done**.
The Salesforce install page showing Installation Complete for the eesel AI package, version 0.2, with a Done button

The package is installed in your org.

{% endstep %} {% step %} ### Authorize the connection Go back to the chat and tell your agent the package is installed. It gives you a **Connect Salesforce** link. Open it and approve the connection in Salesforce. You need permission in Salesforce to authorize a connected app.
The dashboard chat where the user says the package is installed and the agent posts a Connect Salesforce authorization link

The agent posts the authorization link once the package is in.

Your agent only sees what the Salesforce user you authorized can see. Object and field permissions in Salesforce are the real limit, so set them there rather than relying on the instructions you give your agent. This approval is the one step a person has to do in the browser. If an AI agent is doing the setup for you, it hands you the link and waits. {% endstep %} {% step %} ### Confirm the connection Back in the chat, your agent notices the connection on its own. It names your org, checks the connection is healthy, and starts reading your records.
The dashboard chat confirming Salesforce is connected and healthy, naming the org and the number of documents already indexed

The chat confirms Salesforce is connected and healthy.

{% endstep %} {% step %} ### Check what it can read **Integrations > Salesforce** in the left sidebar shows the connection and its sources. Each source has its own toggle, and shows a document count once its sync is done. From a terminal, `eesel status` lists the integration as connected with its source counts.
The connected Salesforce integration page in eesel, listing Cases, Contacts, Accounts and Knowledge Articles as sources with document counts and toggles

Each source has its own toggle and its own document count.

A synced source is a snapshot from the last sync. That is what the agent knows going in. The lookup actions read Salesforce live, which is how it checks one record mid-conversation. {% endstep %} {% step %} ### Create your first automation Connecting alone does not start your agent. It needs an automation first. Tell it what to handle and it creates an [automation](/automations-and-schedules/automations-and-schedules.md). An automation is a Salesforce trigger plus instructions. In the instructions you describe which cases it should handle and what to do with them. It leaves the rest alone.
The dashboard chat where the user asks for an internal case comment with a reply draft on every new case, and the agent reports the new case trigger it enabled

Ask in chat for the automation you want, and the agent builds it.

It shows up on your Automations page, switched on and ready. From a terminal, `eesel automations` lists it as on.
The Automations page in eesel with the A new Case is created rule listed and toggled on

The new automation on your Automations page, switched on.

Open the automation to see or change what it does. This one runs on **A new Case is created** and tells the agent to read the case, draft a reply, and add it as an internal case comment. The trigger event is fixed when the automation is created. So a new automation is required for each different trigger event.
The automation editor showing the name A new Case is created, the Salesforce event it runs on, and the instructions telling the agent to draft a reply and add it as an internal case comment

The automation editor, where you change the instructions it follows.

{% endstep %} {% step %} ### Check automation work in Salesforce Open Salesforce and look at the next case that comes in to see what the automation did there. In this example a customer emailed in about a charge, Salesforce turned the email into a case, and the agent added its drafted reply as an internal case comment for a teammate to review and send.
A Salesforce case created from a customer email, with eesel AI's drafted reply added below it as an internal case comment

The customer email, with the drafted reply under it as an internal case comment.

{% endstep %} {% endstepper %} {% hint style="info" %} **From Claude Code, Cursor or a terminal.** The same setup works as commands: `eesel login`, then `eesel integrations connect salesforce` (a person approves it in the browser; if the receipt has no link, ask `eesel chat` for one), `eesel integrations download start` to copy your sources in, and `eesel automations` to see and switch on the automations. Trigger keys for this page are in the Triggers table below. Full command reference: [The eesel CLI](/apis-and-developer-resources/cli.md). Step-by-step: [Set up from Claude Code or the terminal](/getting-started/set-up-from-a-coding-agent.md). From an MCP client: [Claude Code and MCP clients](/apis-and-developer-resources/claude-code-and-mcp-clients.md). {% endhint %} ## Sources, triggers and actions ### Sources Sources are what it learns from. Toggle each one on or off for each agent from the Salesforce page under Integrations: | Source | Description | When you connect | | ---------------------- | ------------------------------ | -------------------------- | | **Cases** | Your Salesforce support cases | On | | **Contacts** | Your Salesforce contacts | Off until you switch it on | | **Accounts** | Your Salesforce accounts | Off until you switch it on | | **Knowledge Articles** | Your Salesforce Knowledge base | On | Contacts and Accounts still sync when you connect, but the agent can't search them until you switch them on for that agent. They hold names, emails and phone numbers, and an agent that answers your customers could otherwise answer anyone out of your customer directory. This covers searching the synced copy only: the "Look up contact" and "Look up account" actions below still read Salesforce live, so this does not stop the agent from reading a contact on request. If your customers read your Knowledge articles on an Experience Cloud site, enter its address under **Public help center address** in the Knowledge Articles settings. Article links then open the public article instead of its page inside Salesforce, which customers can't open. Links switch over at the next Knowledge Articles sync. eesel indexes what the connecting Salesforce user can see. Salesforce sharing rules don't apply to the indexed copy, so everyone who talks to the agent can reach everything that user could see in the sources you switch on. Connect with a user whose access fits what the agent may read. Every other integration on the agent is also a source. Salesforce answers "what is true for this account" and your help center answers "how does this work", so connect both. See [Integrations](/integrations/overview.md). ### Triggers Triggers are the Salesforce events an automation can run on: | Trigger | Fires when | Best for | Key | | -------------------------------- | ------------------------------------- | ---------------------------------------------------------------- | -------------------------------- | | **A new Case is created** | A case is added, however it got there | Picking up new cases, including ones created from customer email | `salesforce_new_case` | | **A Case is updated** | An existing case is modified | Following a case through its lifecycle | `salesforce_case_updated` | | **A new Lead is created** | A lead is added | Picking up new leads | `salesforce_new_lead` | | **A Lead is updated** | An existing lead is modified | Reacting to a change on a lead | `salesforce_lead_updated` | | **A new Contact is created** | A contact is added | Picking up new contacts | `salesforce_new_contact` | | **A Contact is updated** | An existing contact is modified | Reacting to a change on a contact | `salesforce_contact_updated` | | **A new Account is created** | An account is added | Picking up new accounts | `salesforce_new_account` | | **An Account is updated** | An existing account is modified | Reacting to a change on an account | `salesforce_account_updated` | | **A new Opportunity is created** | An opportunity is added | Picking up new opportunities | `salesforce_new_opportunity` | | **An Opportunity is updated** | An existing opportunity is modified | Reacting to a change on an opportunity | `salesforce_opportunity_updated` | | **A new Campaign is created** | A campaign is added | Picking up new campaigns | `salesforce_new_campaign` | | **A Campaign is updated** | An existing campaign is modified | Reacting to a change on a campaign | `salesforce_campaign_updated` | Each kind of record has its own trigger, and the Salesforce page under Integrations lists the ones your connection offers. One trigger watches one kind of record, so a case automation and a lead automation are two separate automations. The same goes for which records it takes: describe the records you want in the instructions and the agent skips the rest, recording what it passed over on the agent's Activity page. ### Actions Actions are what it can do: | | Action | What it does | | ----------- | --------------------------------- | ------------------------------------------------------------------------------------------------------------------ | | **Respond** | Add case comment | Adds a comment on a case. Mark it internal and only your team sees it. Start here: the agent drafts, a human sends | | | Send email | Sends an email from Salesforce. This one reaches the customer | | **Manage** | Create or update a case | Creates a case, updates its fields, or closes it | | | Create or update a record | Creates or updates a contact, account, lead, opportunity, task or note | | **Look up** | Read case / Search cases | Reads a case in full and finds cases by subject, contact or id | | | Look up contact / Look up account | Finds a contact or an account and reads it live from Salesforce | You stay in control the whole way. Adjust your AI agent's access to knowledge and its action permissions to your liking. See [Actions and Approvals](/instructions-and-memory/actions-and-approvals.md). The lookup actions only read, so they cannot change your data. A write can create a duplicate contact or lead by mistake, so tell the agent to search before it creates. Sending email reaches outside your team. ## Limitations * **The package has to stay installed.** Without the eesel AI package in your org, the connection and the actions stop working. * **It only sees what the authorized user sees.** Object, field and sharing permissions in Salesforce set the real limit. * **Records only, no chat.** Triggers run on Salesforce records. There is no trigger for Salesforce chat or messaging sessions. * **One trigger per automation.** The trigger event is fixed when the automation is created, so each trigger needs its own automation. * **Synced sources are a snapshot.** What it learned from your cases and articles is as of the last sync. The lookup actions read Salesforce live. * **Writes can create duplicates.** Tell your agent to search before it creates a contact or lead. ## Troubleshooting
Connection didn't complete? Check the eesel AI package is installed in your Salesforce org, because the authorization step fails without it. Check you have permission in Salesforce to authorize a connected app. Then complete every prompt: closing the flow partway leaves the connection incomplete. Start the connection again from the chat or from **Integrations** in the left sidebar.
Agent isn't using Salesforce data? Check the source is toggled on under Integrations, and that the record type is within what the authorized Salesforce user can see. Also check your [Instructions](/instructions-and-memory/instructions-and-memory.md) tell it when to look a record up, for example "before answering a billing question, look up the account by the customer's email". A record the lookups cannot find usually does not exist, or is not visible to the authorized user, so check its sharing and field permissions in Salesforce.
Trigger not firing? Check the automation watches the kind of record you expect. A case automation does not fire on a new lead. Then confirm the automation is enabled on your agent's **Automations** page, and that at least one action is enabled.
Actions failing? The eesel AI package has to stay installed in your Salesforce org. If it was removed, the actions stop working. Install it again, then re-authorize from the integration's page.
## FAQ
How do I add an AI agent to Salesforce Service Cloud? Ask your agent in the eesel chat to connect Salesforce, install the eesel AI package it links you to, approve the connection in Salesforce, then ask for an automation. A Salesforce admin has to do the install and the approval in the browser. Your agent gives you the install link first, and once the package is in, the authorization link. A good first automation is "draft a reply as an internal case comment on every new case". Your cases, contacts, accounts and articles then sync on their own. The steps are in [Set up eesel in Salesforce](#set-up-eesel-in-salesforce).
Can the Salesforce AI agent draft replies on a case, or does it send them? Either, depending on how you set it up. It can draft replies as internal case comments that stay inside your team for a person to review and send, or reach the customer directly with **Send email**. It does nothing until you create an automation and enable an action, so you can test with a human in the loop first. To have the agent answer customers itself, connect the helpdesk they write into and let Salesforce supply the account context.
Can I limit which Salesforce cases or records the AI agent handles? Yes. Describe the records you want in the automation's instructions, and the agent skips the rest. The first check reads the case's standard fields, so it can skip by subject, priority, origin, status or email address. Each trigger watches one kind of record, so a case automation and a lead automation are two separate automations. For a hard rule, keep records out of reach with Salesforce permissions, because your agent only sees what the authorized Salesforce user can see.
What is the eesel AI managed package for Salesforce? It is an app you install in your Salesforce org, and Salesforce needs it in place before it allows the connection. Installing it asks you to approve one site, `oracle.eesel.app`, which is eesel. If you uninstall it, the connection stops working.
Is eesel a Salesforce chatbot, or does it replace Agentforce? No, it is not a chat widget inside Salesforce. It is an AI agent that works your cases and records, and you can run it alongside Agentforce or instead of it. If you want a chatbot on your own website, use the eesel [chat bubble](/integrations/knowledge/website.md) and let Salesforce supply the account context.
Can I connect Salesforce to Claude or ChatGPT? Yes, through eesel. Once Salesforce is connected, you can talk to your eesel agent from Claude, ChatGPT or Cursor, and it works with your Salesforce data. See [Use eesel from Claude and other assistants](/apis-and-developer-resources/use-eesel-from-claude-and-other-assistants.md), [Use eesel from ChatGPT](/apis-and-developer-resources/use-eesel-from-chatgpt.md) and [Claude Code and MCP clients](/apis-and-developer-resources/claude-code-and-mcp-clients.md).
What reporting do I get for the Salesforce AI agent? The Activity page lists every case or record your agent handled, and the [Reports](/reports/reports.md) page shows volume, triggers and approvals over time. Each Activity entry links to the record in Salesforce and shows what it searched, each action it took, who approved it, and its reasoning. You can also ask your agent for a written report, or put one on a schedule.
How do I disconnect eesel from Salesforce? Open Salesforce under Integrations in the left sidebar and remove the connection. If you installed the eesel AI managed package, also uninstall it from your Salesforce org to fully revoke access.
## Related pages * [Helpdesk](/use-cases/helpdesk.md) for the job your teammate takes on in a ticket queue * [Automations and Schedules](/automations-and-schedules/automations-and-schedules.md) for triggers, instructions and schedules * [Actions and Approvals](/instructions-and-memory/actions-and-approvals.md) for choosing what your agent may do * [Integrations overview](/integrations/overview.md) for how sources, triggers and actions fit together * [Set up from Claude Code or the terminal](/getting-started/set-up-from-a-coding-agent.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.eesel.ai/integrations/helpdesks/salesforce.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.