Connect HubSpot to Claude

Automatically sync your HubSpot data to Claude using the Windsor.ai connector.

HubSpot’s own Claude connector covers CRM questions and record updates. Windsor.ai puts HubSpot next to Google Ads, GA4, Meta Ads and your billing data, so Claude can show which channels produce closed-won deals rather than leads alone.

Forget CSVs. Stop copy/paste. Connect data in minutes. No code required.

Free forever plan No credit card needed
logo connect
Join over 6,000 businesses that trust Windsor.ai to streamline HubSpot to Claude integration
audi logo
colgate
asics
roche
novo nordisk logo
UBS
unicef
verisure
wpp
club med
skoda
heineken
puma
audi logo
colgate
asics
roche
novo nordisk logo
UBS
unicef
verisure
wpp
club med
skoda
heineken
puma

Which one do I need?

Everything below is about connecting HubSpot to Claude, a separate step that lets you analyze HubSpot data, alone or alongside other sources, from outside the HubSpot UI.

Not to be confused with HubSpot’s built-in AI (Breeze/Copilot). Breeze lives inside HubSpot and is built for in-workflow tasks, like drafting an email or summarizing a contact record.

To help you navigate quicker, here’s what each method allows you to do. “Best for” leads so you can jump straight to the row that matches your situation.

HubSpot’s connector Windsor.ai Automation platforms Custom MCP integration
Best for Everyday CRM queries and record updates Revenue analysis spanning marketing, sales and billing Event-triggered automations Bespoke workflows the others don’t cover
Cross-channel analysis No Yes, 350+ sources No (not built for questions) Only if you build it
Data scope HubSpot only HubSpot plus every other connected source HubSpot plus whatever it connects to Whatever you build
Setup effort Low (native, no code, Super Admin approval) Low (no code) Medium High (developer task)
Write access Yes (create and update, no delete) Read-only Via workflow actions Yes, if you build it
Claude plan needed Paid (Pro, Max, Team, Enterprise) Any plan that supports connectors Depends on the platform Any

Custom MCP integration is last on purpose. It is the heaviest option, a developer task, so it belongs at the end as the fallback for when the other three do not cover your case.

Method 1

HubSpot's official Claude connector

HubSpot launched its connector for Claude in July 2025. As of HubSpot’s August 2026 documentation it reads and writes CRM records, logs activities, analyzes marketing email performance and can create landing pages and blog posts, all from the Claude chat window. It respects the permissions your HubSpot users already have, so nobody sees data through Claude that they could not see in HubSpot itself.

Setup steps

1
Open Connectors in Claude

Sign in to Claude, select your profile icon, then Settings and Connectors.

2
Add HubSpot to your team

Click Browse connectors, select HubSpot, then Add to your team. A HubSpot Super Admin has to approve the connector and choose which data permissions it gets before anyone else can connect.

3
Connect and authorize

Back on the connectors page, click Connect next to HubSpot, log into your HubSpot account, review the permissions and click Connect app.

4
Turn it on in chat

In a Claude chat, open Search and tools and toggle HubSpot on. Works in the Claude web app, desktop and mobile apps.

You need a paid Anthropic plan (Pro, Max, Team or Enterprise) and an active HubSpot user account on any HubSpot tier. Everything runs against HubSpot’s API usage limits. Full details in HubSpot’s connector documentation.

What it can do

  • Read contacts, leads, companies, deals, tickets, custom objects, line items, products, quotes, lists, campaigns, marketing emails, landing pages, blog posts, users and teams
  • Create and update contacts, companies, deals, tickets, custom objects, line items, products, marketing emails, landing pages and blog posts (quotes in beta)
  • Log calls, meetings, notes, tasks and emails against a record, and analyze the full engagement history
  • Chart the results and draft content, with every create or update recorded in HubSpot's audit log

What it can't do

  • Delete records, or create or update more than 10 records in one go
  • Apply your custom validation rules (pipeline stage or association label rules are skipped on writes)
  • Access engagement data when Sensitive Data is turned on in your account
  • See anything outside HubSpot, no ad spend, no website analytics, no revenue data from other tools (see Method 2)

Example prompts

Summarize my active deals by name, amount and stage, sorted by close date.
Which marketing email subject lines generated the most clicks last month?
Log a note on [contact] summarizing today's call and move their deal to the next stage.

Disconnecting: Claude profile icon, then Settings, Connectors, find HubSpot, Manage, Disconnect.

Method 2

Windsor.ai for cross-channel analysis

Windsor.ai covers what HubSpot’s connector does not: it puts HubSpot data next to 350+ other sources (Google Ads, GA4, Meta Ads, LinkedIn Ads, billing systems) so Claude can answer questions that span the whole funnel. Among teams that query HubSpot through Windsor’s Claude connector, 7 in 10 also query Google Ads in the same period, 6 in 10 GA4 and 5 in 10 Meta Ads. The connector is read-only.

Setup takes a few minutes

1
Connect HubSpot to Windsor

Go to onboard.windsor.ai/app/hubspot and authorize with HubSpot's OAuth. Windsor pulls contacts, companies, deals, owners, marketing emails, forms, lists, calls, meetings and tickets, with custom properties detected automatically.

2
Add the Windsor.ai connector to Claude

In Claude, open Customize, then Connectors, search for Windsor and click Add. Sign in to your Windsor.ai account when prompted, and the connector appears in your Connectors list. You can also open the Windsor.ai listing in the Claude directory (https://claude.ai/directory/360c0c31-4bb6-42ca-8e50-5da0a100a68e) and click Connect.

3
Ask questions

Start a new chat and ask about your pipeline, lead sources or email campaigns in plain language, alone or next to your ad and analytics data.

No code, API calls or custom pipelines required. Runs on Windsor MCP for Claude. Screenshots for each step are in the setup guide.

See it in your own HubSpot data

A walkthrough of connecting HubSpot to Claude through Windsor.ai.

Skip exports and dashboard building, get clean, ready-to-use answers straight in Claude
Puts HubSpot next to Google Ads, GA4 and Meta Ads so Claude can trace spend to closed-won deals in one conversation
Groups contacts by original source and lifecycle stage to show which channels convert
Reads 1,100+ fields across contacts, deals, companies, emails, forms and tickets, custom properties included

What it can do

  • Query HubSpot alongside Google Ads, GA4, Meta Ads, LinkedIn Ads and billing data in one conversation
  • Pull 1,100+ fields across contacts, deals, companies, owners, marketing emails, forms, lists, calls, meetings and tickets, custom properties included, queried live
  • Go back as far as your HubSpot history allows, teams ask about anything from today to the last two years
  • Answer questions HubSpot's own connector cannot, such as which ad campaigns produced closed-won deals rather than leads
  • Connect a second CRM such as Salesforce in the same chat

What it can't do

  • Write back to HubSpot, the connector is read-only (use Method 1 to create or update records)
  • Replace the per-contact engagement timeline, that is the official connector's strength

What data is available

Ordered by how many teams query each object through Claude. Contacts and deals lead by a wide margin; marketing emails, companies, owners and forms follow.

HubSpot object Key fields What Claude can do with it
Contacts contact_hs_analytics_source, contact_lifecyclestage, contact_createdate, contact_count, contact_email, contact_hs_lead_status, contact_first_conversion_event_name Lead source and lifecycle funnels, new contacts by channel, lead status audits
Deals deal_dealstage, deal_amount, deal_dealname, deal_pipeline, deal_createdate, deal_closedate, deal_hs_is_closed_won, deal_hs_analytics_source Pipeline value by stage, win rates, source-to-revenue, deal aging
Marketing emails campaignemails_name, campaignemails_subject, campaignemails_stats_counters_sent, campaignemails_stats_counters_open, campaignemails_stats_counters_click, campaignemails_stats_ratios_openratio, campaignemails_stats_ratios_clickratio Subject line comparison, open and click ratios, unsubscribe trends
Companies company_name, company_domain, company_industry, company_annualrevenue, company_numberofemployees Account segmentation, fit scoring, industry breakdowns
Owners and pipelines owner_first_name, owner_last_name, owner_email, deal_hubspot_owner_id, pipeline_name, pipeline_stage_name, pipeline_stage_probability Rep performance, weighted forecasts, stage definitions
Forms and lists forms_name, formsubmissions_pagetitle, formsubmissions_submittedat, lists_name, lists_size Submissions per page over time, list growth
Engagements and tickets calls_hs_call_duration, meetings_hs_meeting_title, tickets_createdate, tickets_closed_date Activity volume per rep, ticket resolution times

Queried live against HubSpot’s API, so there is no fixed sync schedule or cache delay. Full list of 1,100+ fields in the HubSpot field reference.

Example prompts

Lead sources and lifecycle

Show contact_count by contact_hs_analytics_source and contact_lifecyclestage for contacts created in the last 30 days. Which sources produce customers, and which stop at lead?
Break down new contacts by contact_hs_analytics_source_data_1 for the last 90 days, month by month. Which campaigns or referrers are growing?

Pipeline and deals

Sum deal_amount and deal_count by deal_dealstage and deal_pipeline for open deals. Which stage holds the most value, and which deals have sat there longest by deal_createdate?
For deals with deal_hs_is_closed_won true in the last 2 years, group deal_amount by deal_hs_analytics_source. Which original source produces the most closed-won revenue, and what is the average deal size per source?

Sales reps

Rank deal owners by closed-won deal_amount for the last 90 days using owner_first_name and owner_last_name. Show deal_count and win rate per owner, and where each owner's lost deals stalled.

Marketing emails

List campaign emails from the last 30 days with campaignemails_subject, sent, open ratio, click ratio and unsubscribes. Which subject lines drove the highest click ratio, and what do they have in common?

Forms and lists

Count form submissions by formsubmissions_pagetitle for the last 30 days and compare with the previous 30. Which pages are gaining or losing submissions?

Cross-channel

Join Google Ads spend by campaign with HubSpot contacts by contact_hs_analytics_source_data_1 for the last 30 days. Show cost per new contact and cost per closed-won deal per campaign.
Compare Meta Ads and Google Ads for the last 90 days on spend, contacts created in HubSpot and closed-won deal_amount by source. Which channel produces the cheaper closed-won deal?
Using GA4 sessions by source and HubSpot contacts by contact_hs_analytics_source, show the session-to-contact conversion rate per channel for the last 30 days.

Cross-channel questions this unlocks

  • Which Google Ads keywords and campaigns are driving SQLs and closed-won deals
  • Which LinkedIn Ads or Meta Ads campaigns produce deals rather than leads that never close
  • How GA4 behavior on high-value leads lines up with deal progression
  • Whether pipeline value reconciles with revenue in your billing system

What sales and marketing teams use this for

Revenue forecasting on demand

Instead of waiting for the weekly pipeline review, ask Claude for a weighted forecast at any moment, using live deal stages, stage probabilities and expected close dates from HubSpot.

Pre-meeting briefs

Before a client call, ask Claude to summarize a contact's HubSpot history: company details, open deals, logged calls and meetings, and the last marketing email they engaged with.

Marketing and sales alignment

Combine HubSpot with Google Ads or LinkedIn Ads in the same Claude chat to ask which paid campaigns generate contacts that close, and at what deal size.

Onboarding and ramp analysis

Compare the pipeline progression of top performers against newer reps to see where newer reps lose deals that experienced reps close.

CRM hygiene reports

Ask Claude for contacts with missing required fields, deals with no close date, or companies with no associated contacts, and run a targeted cleanup instead of a quarterly audit.

Method 3

Automation platforms

If what you need is event-triggered automation, such as “when a deal closes, do X”, rather than conversational questions, workflow platforms like Zapier, Make or n8n can connect HubSpot and Claude for that. It is a different job from Methods 1 and 2: those answer questions about your data, this triggers actions off HubSpot events.

What it can do

  • Trigger actions automatically when HubSpot events happen, such as drafting a follow-up when a deal closes
  • Connect HubSpot to Claude as one step in a broader workflow spanning other tools

What it can't do

  • Answer conversational questions about your data, it is not built for that (see Methods 1 and 2)
  • Run itself, broken runs fail without telling anyone and the fix is on you

A plain HubSpot export to CSV also works for one-off analysis, but you lose the relationships between objects, so deals arrive disconnected from contacts. Fine for a single quick question, not for anything recurring.

Method 4

Custom MCP integration on HubSpot's API

If none of the above cover a workflow you need (a specific internal object, a non-standard permission model, an action HubSpot has not exposed yet), you can build your own MCP server against HubSpot’s API and expose it to Claude directly.

This is the heaviest route here: you host the MCP server and handle auth, token refresh and HubSpot’s rate limits yourself rather than doing a no-code setup.

What it can do

  • Whatever you build, full flexibility over which HubSpot objects, actions and permissions are exposed to Claude
  • Cover workflows that the official connector or Windsor.ai do not support

What it can't do

  • Work without developer effort, there is no no-code path here
  • Avoid ongoing maintenance, HubSpot API versions and access tokens rotate

Reach for this only when Methods 1 to 3 do not cover your case. It carries build and maintenance cost that the other three do not.

Conclusions: which method wins for your use case

Your situation Recommended Why
“I need HubSpot data alongside ad spend, GA4 or billing data” Windsor.ai Only method here that spans 350+ sources beyond HubSpot
“I need to know which campaigns produce closed-won deals, and not only leads” Windsor.ai Cross-channel analysis is what the connector is built for
“I want HubSpot and Salesforce, or another CRM, in the same chat” Windsor.ai Both connect through one Windsor.ai connector
“I just want to ask Claude about deals and contacts, no dev work” HubSpot’s official connector Native, no code, respects your existing HubSpot permissions
“I want to create or update HubSpot records from chat” HubSpot’s official connector Only no-code method here with write access
“I want something to happen automatically when a deal closes” Automation platforms Built for event-triggered actions, not conversational questions
“None of the above cover what I need, and I have dev resources” Custom MCP integration Full flexibility, at the cost of build and maintenance effort

FAQs

What are the main ways to connect HubSpot to Claude?

There are four options: HubSpot’s official Claude connector (native, no code, reads and writes CRM records, respects your HubSpot permissions), Windsor.ai (no code, joins HubSpot with 350+ other sources for cross-channel analysis, read-only), automation platforms (for event-triggered actions rather than questions), and a custom MCP server built against HubSpot’s API (full flexibility, requires developer resources). See the comparison above for which fits your case.

Tired of manual HubSpot data exports? Try Windsor.ai today to automate your reporting

Access your data from various sources in one place. Get started for free.
g logo
fb logo
big query data
youtube logo
power logo
looker logo