Find stalled deals
Which open opportunities over $25,000 have stayed in the same stage for more than 30 days? Group them by owner and tell me which ones to push this week.
Your AI tool never touches Salesforce directly. It asks Coupler.io, which runs the calculation on your synced CRM data and returns only the computed result.
Shape your Salesforce data for AI before anyone asks a question
Query one prepared dataset from the AI client your team already uses
Move from an answer to a decision while the context is still fresh
I'm not getting AI's best guess. Coupler.io does that actual math. Claude just helps me ask the right question and understand the results conversationally
Copy a prompt into your AI tool and swap in your own dates and quota.
Which open opportunities over $25,000 have stayed in the same stage for more than 30 days? Group them by owner and tell me which ones to push this week.
Our win rate fell this quarter. Compare it with the last two quarters by lead source, segment, and deal size, and point to the biggest driver.
Is our Q4 pipeline coverage above 3x quota? Show weighted and unweighted pipeline by region and flag every region below target.
List opportunities whose close date moved more than twice in the last 60 days. Which reps have the most, and what's the total amount at risk?
Show win rate, average deal size, and median sales cycle for each rep over the last 6 months. Leave out reps with fewer than 10 closed deals.
How many days do opportunities spend in each stage this year, on median? Where is stage-to-stage conversion lowest?
Which campaigns sourced the most closed-won revenue in the last 90 days? Put it next to Google Ads and LinkedIn Ads spend and rank campaigns by return.
Break down closed-lost opportunities from last quarter by loss reason and competitor. Which reason grew fastest compared with the quarter before?
Write a one-page pipeline summary for leadership: commit versus best case by forecast category, the five largest deals at risk, and what changed since last week.
A skill is a written procedure your AI follows step by step. This week's pipeline review checks exactly what last week's did.
Analyzes pipeline and CRM data from Salesforce, HubSpot, Pipedrive, and more. Covers win rate, deal velocity, stage conversion, and rep performance. Works as a standalone skill or combined with marketing-analytics for full-funnel orchestration.
Turns raw analysis into a clean, decision-ready report. Runs validation on figures and claims before delivery so you can share it with stakeholders without a second pass. Outputs markdown, summarize-ready text, or formatted tables.
Building your own AI agents, or automating in n8n or Make? The Custom MCP destination turns a Salesforce data flow into an endpoint any MCP-compatible client can query.
Sign in to your Salesforce org and set a refresh schedule.
Coupler.io generates a token and a server URL.
Claude Code, Cursor, and VS Code accept a remote server, as does any other client that supports the Model Context Protocol.
{
"mcpServers": {
"coupler": {
"url": "https://mcp.coupler.io/mcp/10146",
"headers": {
"Authorization": "Bearer YOUR_TOKEN"
}
}
}
}
Salesforce's hosted MCP servers became generally available in April 2026 as part of the Headless 360 push. They give AI agents live, permission-aware access to records, with read-only and write-enabled server options. Admins can also add custom tools built on flows, an Apex class, or named queries. For Salesforce developers, the Salesforce DX MCP server covers metadata and testing work inside the IDE. Coupler.io focuses on analysis: it syncs your Salesforce data and calculates the answers your AI tool explains.
Connect Salesforce to your AI tool through Coupler.io and start with the deals that stopped moving. You don't need an External Client App or a server to host.
Model Context Protocol (MCP) is an open standard that lets AI tools read data from other systems. A Salesforce MCP server connects your Salesforce org to an AI assistant, and you can ask about opportunities, accounts or pipeline in plain language from Anthropic's Claude AI, ChatGPT or Gemini. Coupler.io runs it as a hosted Salesforce MCP server connector. Sign in to Salesforce, pick your AI tool and start asking.
Salesforce offers two options. With the Salesforce hosted MCP servers, an admin activates servers such as SObject Reads, sets up an External Client App and connects AI clients through per-user authentication. These need Enterprise Edition or above. The Salesforce DX MCP server is a developer tool. It runs locally against orgs you've authorized with the Salesforce CLI and helps you deploy metadata or run Apex tests from VS Code, Cursor or Agentforce Vibes. Coupler.io's Salesforce MCP connector is built for analysis. It works with Standard and Professional editions too, calculates in SQL and blends Salesforce with 400+ other sources.
Coupler.io does the arithmetic and the AI does the explaining. The Analytical Engine runs SQL on your synced Salesforce data and returns computed results, which the LLM then puts into words. You can open the query behind any figure. Saved definitions, like how your team weights each forecast category, keep the model's reading in line with your own reports.
The connector reads from Salesforce and doesn't write back. You still update opportunities or reassign leads in Salesforce itself. Answers reflect your last sync, and the AI sees only the Salesforce objects and fields in your data flow. Report imports support tabular reports and return the first 2,000 rows. For bigger datasets, pull Objects or write SOQL queries. Connecting needs the REST API enabled for your org, and both production and sandbox orgs work.
No. Your AI tool never gets your Salesforce credentials or direct access to your org. When you ask a question, the Analytical Engine runs the query, and the model receives only column names, data types, a sample of rows, your saved context and the calculated result. Your full CRM data stays in Coupler.io, which is SOC 2 Type II certified and compliant with GDPR, HIPAA and DORA. Credentials are encrypted with AES-256, and data in transit is protected with TLS.
