Destinations
Data sources
AI insights
About us
Careers
Security
Customer reviews
Contact us
Affiliate program
Solution partners
Template gallery
Looker Studio templates
Power BI templates
Excel templates
Google Sheets templates
Tableau templates
Facebook Ads templates
Google Ads templates
Data fields & Metrics
AI prompt library & Guides
Product documentation
API documentation
Case studies
Blog
Data models
Windsor vs Supermetrics
Windsor vs Funnel.io
Windsor vs Fivetran
Windsor vs Rivery
Windsor vs Portermetrics
Windsor vs Coupler.ioConnect Salesforce to ChatGPT
Automatically sync your Salesforce data to ChatGPT using the Windsor.ai Salesforce ChatGPT connector.
Connect Salesforce to ChatGPT to analyze pipeline performance and get AI-driven recommendations that sharpen sales strategy and speed up revenue growth.
Forget CSVs. Stop copy/paste. Connect data in under a minute. No code required.
Free forever plan
Which one do I need?
Salesforce now has two native apps for ChatGPT, alongside Windsor.ai’s cross-channel connector, manual exports, and custom builds. Which one you need depends on your Salesforce edition, whether you need other data sources alongside your CRM, and how much setup effort you want to take on.
| Salesforce’s native apps | Windsor.ai | Manual CSV export | Custom GPT Actions | |
|---|---|---|---|---|
| Best for | Enterprise Edition orgs, or Agentforce customers wanting a packaged sales app | Recurring analysis spanning Salesforce plus ad platforms, web analytics, and other channels | A one-off report with no ongoing connection needed | Bespoke workflows the other three don’t cover |
| Built for conversational Q&A | Yes | Yes | Yes, but on a static snapshot | Yes, if you build it that way |
| Cross-channel analysis | No, Salesforce only | Yes: Salesforce + 350+ sources | No | Only if you build it |
| Connection type | Live: direct to your org | Live: direct to your org, no re-referencing | Static snapshot from export time | Live, if you build it that way |
| Setup effort | Low to medium; higher for Hosted MCP Servers | Low (no code) | Very low, but manual every time | High (developer task) |
| Write access | Depends on the app; Hosted MCP Servers can expose Apex actions | Yes, on Opportunity, Lead, Contact, Account, Case, or custom object fields | No | If you build it that way |
| Requires a paid add-on or edition | Yes: an Agentforce add-on, or Enterprise Edition and above | No | No | No, beyond a paid ChatGPT plan to build a Custom GPT |
Salesforce's own native apps
Salesforce has two native ways into ChatGPT, both built and maintained by Salesforce itself, not a third party. The Agentforce Sales app, launched in ChatGPT’s app store as part of Salesforce’s expanded OpenAI partnership, gives authorized users natural-language access to leads, opportunities, and account records. It requires an Agentforce for Sales Add-on or Agentforce 1 Edition license, on top of your existing Salesforce subscription.
Hosted MCP Servers, generally available on Enterprise Edition orgs and above, are a more flexible option: Salesforce hosts the MCP server for you and can expose not just standard CRM data but flows, Apex actions, and custom queries, connected through ChatGPT’s Developer Mode.
What it can do
- Use Salesforce's own supported, officially maintained access path, with no third-party service involved
- Expose custom flows, Apex actions, and bespoke queries through Hosted MCP Servers, not just standard CRM fields
- Get enterprise-grade permission enforcement, since access follows your org's existing Salesforce permissions
What it can't do
- Blend Salesforce with ad platforms, web analytics, or other channels, since both apps only see Salesforce data (see Method 2)
- Use the Agentforce Sales app without an Agentforce add-on license, an extra cost on top of your Salesforce subscription
- Use Hosted MCP Servers without an Enterprise Edition org or above, or without ChatGPT Developer Mode
Windsor.ai for cross-channel analysis
Windsor connects to the Salesforce API and pulls your core CRM objects: leads, opportunities, accounts, contacts, and activities. Standard objects are included by default, and any custom objects or custom fields in your org are detected and mapped automatically when you authorize the connection. Because Windsor already connects 350+ sources, you can blend Salesforce with Google Ads, Meta Ads, or GA4 in the same conversation, something neither of Salesforce’s own native apps supports.
Pardot, Salesforce’s marketing automation product, is a separate Windsor connector rather than part of the Salesforce CRM connector, so connect it separately if you also want Pardot campaign data alongside your CRM data.
What it can do
- Read and write, since Windsor MCP supports updating field values on an existing Opportunity, Lead, Contact, Account, Case, or custom object, by record ID
- Blend Salesforce with 350+ other sources, including Google Ads, Meta Ads, and GA4, in one conversation
- Set up in under a minute with a single OAuth authorization, no code, no API credentials to manage
What it can't do
- Create or delete Salesforce records, since the current write action only updates fields on an existing record
- Access flows or Apex actions the way Salesforce's own Hosted MCP Servers can (see Method 1)
Setup takes under a minute
Go to onboard.windsor.ai/app/salesforce and authenticate via OAuth using your Salesforce credentials. Select the account(s) you want to pull data from; standard and custom objects are detected and mapped automatically.
Open the Windsor.ai app page in ChatGPT and click Connect, then complete the OAuth flow to grant access.
Open a new chat and ask about your pipeline, leads, or accounts. Windsor pulls live from the Salesforce API on every question, not a cached snapshot.
The fields that matter most for Salesforce
| Field | What it tells you |
opportunity_amount |
The monetary value of an open or closed opportunity |
opportunity_stage_name |
The current stage of the deal in your sales process |
opportunity_is_won |
Whether the opportunity was closed as a win |
opportunity_last_activity_date |
The most recent date any activity was logged on the opportunity |
lead_source |
The channel or campaign that generated the lead |
lead_is_converted |
Whether the lead has been converted to an opportunity |
account_industry |
The industry classification of the account |
campaign_expected_revenue |
The forecasted revenue from the campaign |
campaign_number_of_converted_leads |
Leads from the campaign that converted to opportunities |
Windsor exposes 30+ metrics and 265+ dimensions from Salesforce in total. See the full Salesforce field reference.
How to connect Windsor.ai to ChatGPT
Connect 350+ data sources, including Salesforce, to ChatGPT in 3 clicks with the Windsor.ai app.
Example prompts
Pipeline health
Summarize my Salesforce pipeline by stage for the last 30 days.
For each stage, show total number of open opportunities, total pipeline value, average time deals have spent in this stage, and conversion rate to the next stage.
Which stage has the lowest conversion rate and the longest average duration? Flag it as the primary bottleneck.
Win and loss patterns
Compare won and lost opportunities in Salesforce for the last 30 days.
For each outcome, show the breakdown by lead source, industry, deal size range, and assigned rep.
Which combination of factors is most strongly associated with a closed-won outcome? Which segments have a win rate below 20%?
At-risk deals
Show me all open opportunities in Salesforce with no activity logged in the last 21 days.
For each, include the opportunity name, assigned rep, current stage and days in that stage, pipeline value, and last activity date.
Sort by pipeline value, highest first. Which of these represent the most immediate revenue risk?
Rep performance
Analyze sales rep performance in Salesforce for the last 30 days.
For each rep, show total closed-won revenue, number of deals closed, average deal size, average days from creation to close, and win rate.
Which reps have the best combination of win rate and deal velocity?
Revenue momentum
Analyze my closed-won revenue in Salesforce month by month for the last 6 months.
For each month, show total closed-won revenue, number of deals closed, average deal size, and month-over-month change.
Which months show the biggest swings, and what factors appear to be driving the changes?
For Salesforce admins: Windsor’s connection follows the same OAuth permission model as any other connected app. Admins can see and revoke it from Setup like any other authorized app, and it doesn’t require changing field-level security or sharing rules to work. Custom objects and custom fields specific to your org are detected automatically, so there’s no manual field mapping to maintain as your schema changes.
Manual CSV export
Salesforce Reports can export directly to CSV, which you can then upload into a ChatGPT conversation. It works with no third-party tool and no setup, but the data is a static snapshot from the moment you export it, and object relationships between leads, opportunities, and accounts are lost in the flat file, making cross-object analysis difficult.
What it can do
- Get started immediately with no account setup or authorization
- Analyze a single report without connecting anything to your org
What it can't do
- See changes made in Salesforce after the export, since ChatGPT only has the snapshot you uploaded (see Method 2)
- Analyze relationships across leads, opportunities, and accounts together, since those are flattened into separate files
Custom GPT Actions via the Salesforce REST API
Teams with development resources can register a Connected App in Salesforce, configure OAuth scopes, and build a Custom GPT Action with an OpenAPI schema that maps to the Salesforce REST API directly. This gives full control over which objects and fields are exposed, without going through Salesforce’s own Hosted MCP Servers or Windsor.
For orgs on Enterprise Edition or above, Salesforce’s own Hosted MCP Servers (see Method 1) now cover much of the same ground with less custom code, since Salesforce hosts and secures the server for you. This route still makes sense below Enterprise Edition, or when you need logic that neither native app nor Windsor expose.
What it can do
- Whatever you build: full control over which objects, fields, and actions are exposed
- Cover workflows none of the other three methods support
What it can't do
- Work without developer effort or ongoing maintenance as your Salesforce API version or org configuration changes
- Avoid needing a paid ChatGPT plan (Plus, Pro, Business, Enterprise, or Edu) to build a Custom GPT at all
Conclusions: which method wins for your use case
| Your situation | Recommended | Why |
|---|---|---|
| “I need Salesforce data alongside ad spend, GA4, or other channels” | Windsor.ai | Only method here that spans 350+ sources beyond Salesforce |
| “I want ChatGPT to update Salesforce records” | Windsor.ai | Simplest write-capable path; Salesforce’s own apps focus mainly on read access |
| “I’m on Enterprise Edition and want Salesforce’s own supported access” | Salesforce’s native apps | Officially built and maintained by Salesforce itself |
| “I just need one report, right now, no setup” | Manual CSV export | No account or authorization needed |
| “None of the above cover what I need, and I have dev resources” | Custom GPT Actions | Full flexibility, at the cost of build and maintenance effort |
FAQs
Is Windsor's Salesforce connector a ChatGPT plugin?
No. ChatGPT retired its original plugin system. Windsor connects through ChatGPT’s current native app and connector model instead, the same mechanism Salesforce’s own Agentforce Sales app uses.
Can ChatGPT update Salesforce records, not just analyze them?
Yes. The Windsor connector isn’t read-only: once it’s connected, you can ask ChatGPT to update field values on an existing Salesforce record, an Opportunity, Lead, Contact, Account, Case, or a custom object. Only the fields you specify are changed; the record itself isn’t created or deleted, and you can review the update before it’s applied.
Does Windsor support Pardot data too?
Yes, but as a separate connection. Pardot, Salesforce’s marketing automation product, is its own Windsor connector distinct from the core Salesforce CRM connector. Connect both if you want Pardot campaign and email data alongside your CRM data in the same ChatGPT conversation.
Can I connect multiple Salesforce orgs to Windsor?
Yes. Multiple orgs can be connected to Windsor depending on your plan, and all of them are queryable within the same ChatGPT conversation. This is particularly useful for agencies or enterprise teams managing more than one Salesforce environment.
Does the connector work with custom fields and custom objects?
Yes. Windsor automatically detects and maps custom fields and custom objects the moment you authorize your Salesforce account via OAuth. Your org-specific data structure is available in ChatGPT without any manual selection, mapping, or configuration.
How fresh is the Salesforce data ChatGPT sees through Windsor?
Windsor pulls live from the Salesforce API each time you ask a question in ChatGPT. You’re not working from a cached snapshot or a scheduled export from hours ago; every analysis reflects the current state of your pipeline, leads, and account data at that moment.
Is Windsor MCP free? What plan do I need?
Windsor has a free forever plan, no credit card required. Paid plans start at $19/month for higher data volume and additional sources. See the pricing page.
Do you have helpful links to get started with integrating Salesforce and ChatGPT using Windsor.ai?
Yes, we have helpful resources to help you get started. You can explore our official documentation and available data fields to get started:
Tired of manual Salesforce data exports? Try Windsor.ai today to automate your reporting
