Salesforce MCP server: ask questions, get governed answers

Salesforce AI analytics: what you can ask

Typical questions from sales leaders and RevOps. Each answer shows the definition, source and query behind it.

• What is the open pipeline by stage and forecast category this quarter?
• Which opportunities are past their close date and still open?
• Compare closed-won amount by region this quarter versus last quarter.
• Who are the top 10 reps by closed-won amount this year?
• How many leads were created last month by lead source?
• Which accounts have the most open cases right now?

Worked example: a governed Salesforce answer

One question, and everything the answer shows so you can check it. The figures are invented sample data.

SELECT ForecastCategoryName, COUNT(Id) deals, SUM(Amount) amount
FROM Opportunity
WHERE IsClosed = false
  AND CloseDate = THIS_FISCAL_QUARTER
GROUP BY ForecastCategoryName

Common pitfalls when asking Salesforce questions

Each one is settled once in an approved definition, so every person and every assistant gets the same answer.

• Current stage versus stage history

  StageName shows where a deal is today. A question such as "deals that reached Proposal last quarter" needs opportunity history instead. The definition states which one a metric uses.

• Fiscal and calendar quarters

  THIS_QUARTER and THIS_FISCAL_QUARTER return different deals when your fiscal year does not start in January. The definition fixes the calendar once, for every question.

• Mixed currencies

  With multiple currencies enabled, each opportunity's Amount is in its own currency. The definition says which currency to report in and which exchange rates apply, so totals are never a mix.

How Nexatron connects to Salesforce

Whose permissions

Each person can connect their own Salesforce account, so answers return only the records that person can already see in Salesforce. An administrator can instead use one shared connection for the workspace.

Read and write

Questions are read-only. A small set of Salesforce actions also exists; each one is previewed first, runs only after explicit confirmation by a permitted person, and can be switched off by your administrator.

Setup

Sign-in happens on the vendor's own OAuth consent screen; Nexatron never sees a password. Steps: choose the source → check prerequisites → enter connection details → approve access → verify → ask. No data migration and no pipeline to build: data is queried where it lives.

Salesforce MCP connector for Claude, ChatGPT and Copilot

Nexatron's hosted MCP server is a governed Salesforce MCP connector: Claude, ChatGPT and Microsoft Copilot ask questions of your Salesforce data through it. Each person signs in as themselves, so the assistant sees only what they may see.

Connect Salesforce to Claude, ChatGPT or Copilot

1. Connect Salesforce to your Nexatron workspace.
2. In Claude, add Nexatron from Claude's connector directory. In ChatGPT or Copilot, add a custom MCP connector with the server URL https://api.nexatron.io/api/v1/mcp.
3. Sign in with your own Nexatron identity (OAuth 2.1), then ask your Salesforce questions.

Salesforce connector FAQ

Can I ask Salesforce questions without writing queries?

Yes. The connector is available now. Ask in Teams, Slack, the Nexatron web app, or an AI assistant connected through Nexatron's MCP server.

Does Nexatron respect Salesforce permissions?

Yes. With a personal connection, answers return only the Salesforce records that person can already see.

Can Nexatron change data in Salesforce?

Only through optional actions that a permitted person previews and confirms. Asking a question never changes Salesforce data.

What do I need to connect Salesforce?

A user who can approve access on the vendor's own sign-in screen.

Is there a Salesforce MCP server for Claude, ChatGPT or Copilot?

Yes: Nexatron's hosted MCP server, once Salesforce is connected to your workspace. It is Nexatron's server, not one published by Salesforce, and the assistant gets the same governed answer as the Nexatron app.

How is asking Nexatron different from building a Salesforce dashboard?

A Salesforce dashboard answers the questions it was built for. Nexatron answers the follow-up question, applies the same approved definition every time, and shows the SOQL behind the number so you can check it against the report you trust. Keep dashboards for the views you check every day.

Do I need to copy Salesforce data into a warehouse for AI analytics?

No. Nexatron queries Salesforce through its API when a question is asked, so there is no copy to build or keep in sync. If you already have Salesforce data in a warehouse, you can connect that too and choose which source each definition uses.

Can a Salesforce answer include data from other systems?

Yes, when those systems are connected and federation is on for your workspace, as it is by default for new workspaces. Nexatron then pulls only the fields the question needs from each source and joins them inside your tenant, for example closed-won amount next to invoiced revenue from your ERP. The answer lists every source it used.