Claude Code with HubSpot
HubSpot has AI features. Content assistant. Conversation intelligence. Predictive lead scoring. But none of it knows YOUR sales process.
It doesn't know that enterprise deals take 90 days and need three demos. It doesn't know that you always discount if they commit to annual. It doesn't know that prospects from referrals close at 60% while cold outreach closes at 8%.
HubSpot's AI treats every lead the same because it only sees the fields you filled in. It doesn't understand your context.
Claude Code could. But it can't access your HubSpot data. Every time you ask Claude to draft a follow-up email or analyze your pipeline, you're re-explaining information that already exists in your CRM.
Why HubSpot Needs AI Memory
Your CRM contains everything you know about your sales process. Contact history. Deal stages. Email sequences. Win/loss patterns. But that data is trapped in HubSpot's interface.
When you draft emails, Claude doesn't know:
- The prospect's industry, company size, or pain points
- What stage they're at in your pipeline
- Previous conversations or objections they raised
- Which email templates work for their segment
- Your standard pricing, objection handling, or close tactics
When you analyze deals, Claude doesn't know:
- Your pipeline stages and what each means
- Which deal characteristics predict closes
- Where deals typically get stuck
- Your team's activity benchmarks
HubSpot holds the data. Claude has the intelligence. They just can't talk to each other.
How to Connect HubSpot to Claude Code
You're not building a custom integration. You're creating a markdown-based context layer that makes your HubSpot data readable by Claude.
The architecture:
- Export key HubSpot data — Contact lists, deal pipelines, email templates, sales playbooks
- Convert to markdown and store in an Obsidian vault
- Create a CLAUDE.md file that documents your sales process and data structure
- Let Claude Code read the vault whenever it needs CRM context
HubSpot stays your CRM. The vault becomes your AI's memory. Claude bridges them.
What to Export from HubSpot
Don't export everything. Export the context Claude needs to understand your sales process:
- Contact segments — Active prospects, past customers, referral sources with key details
- Deal pipeline structure — Stage definitions, average time in each stage, what moves deals forward
- Email templates — Your best-performing sequences for each stage and segment
- Sales playbooks — How you qualify, demo, handle objections, close
- Product/service descriptions — What you sell, pricing tiers, typical use cases
- Win/loss patterns — What closes deals, what kills them
Export as CSV for lists, copy-paste for templates, and document your process in markdown.
Exporting Contact Lists
In HubSpot, create filtered views of your key segments:
- Active opportunities
- High-value prospects
- Referral sources
- Past customers
Export each as CSV. Convert to markdown tables with key columns:
| Company | Contact | Industry | Stage | Value | Last Touch | Notes |
|---------|---------|----------|-------|-------|------------|-------|
| Acme Corp | Jane Smith | Manufacturing | Demo Scheduled | $50K | 2026-01-20 | Referred by John. Needs Q1 start. |
| Beta Inc | Mike Chen | SaaS | Proposal Sent | $30K | 2026-01-18 | Price-sensitive. Asked about monthly. |
Don't export your entire database. Export active deals and reference contacts. Update monthly or when significant changes happen.
Documenting Your Pipeline
HubSpot shows deal stages. It doesn't explain what they mean or how to move deals through them.
Create a markdown file that documents:
- Each pipeline stage
- What qualifies a deal for that stage
- Typical time in stage
- What actions move deals forward
- Common sticking points
Example:
## Pipeline Stages
### 1. Qualified Lead
- **Definition:** Expressed interest, matches ICP, has budget authority
- **Avg time:** 7 days
- **Next action:** Schedule discovery call
- **Exit criteria:** Confirmed pain point + budget
### 2. Discovery Call
- **Definition:** Initial conversation completed
- **Avg time:** 3 days
- **Next action:** Send proposal or schedule demo
- **Exit criteria:** They agree to next step or they decline
### 3. Demo Scheduled
- **Definition:** They agreed to product demonstration
- **Avg time:** 10 days
- **Next action:** Complete demo, send follow-up
- **Red flags:** Rescheduling multiple times, vague about timeline
This context helps Claude understand where deals are and what to do next.
Exporting Email Templates
Your HubSpot templates contain proven language that closes deals. Export them so Claude can learn your voice and structure.
For each template, create a markdown file with:
- Template name and purpose
- When to use it (stage, segment, scenario)
- The template itself
- Performance notes (open rate, response rate)
Example:
---
template: post-demo-followup
stage: Demo Complete
segment: Enterprise
response-rate: 68%
---
# Post-Demo Follow-Up (Enterprise)
**Subject:** Next steps after today's demo
Hi [First Name],
Thanks for taking the time to walk through [Product] with me today. Based on our conversation, it sounds like [specific pain point] is the top priority.
Here's what I recommend:
[Customized next steps based on demo discussion]
I've attached a proposal that includes [specific features discussed] with pricing for [size/tier discussed].
How does [specific date one week out] look for a follow-up call to answer any questions?
[Your Name]
Claude can reference these when drafting emails. It learns what works instead of writing generic AI slop.
Documenting Your Sales Playbook
Your playbook is your accumulated sales knowledge. How you qualify. How you handle objections. How you close. It's probably in your head, scattered across training docs, or buried in HubSpot notes.
Export it to markdown:
- Lead qualification criteria — What makes a good prospect
- Discovery questions — What you always ask and why
- Objection handling — Common objections and proven responses
- Closing tactics — What works at different price points
- Pricing strategy — When to discount, when to hold firm, how to present tiers
This gives Claude the sales intelligence to help you think through deals, not just draft generic emails.
Setting Up Your Vault Structure
Your Obsidian vault needs clear organization so Claude can find CRM context fast.
Example structure:
/CRM/
/Contacts/
active-opportunities.md
past-customers.md
referral-sources.md
/Pipeline/
pipeline-structure.md
deal-stage-definitions.md
win-loss-analysis.md
/Templates/
/Email/
discovery-outreach.md
post-demo-followup.md
proposal-delivery.md
/Proposals/
enterprise-proposal-template.md
smb-proposal-template.md
/Playbook/
lead-qualification.md
objection-handling.md
pricing-strategy.md
closing-tactics.md
/Product/
service-descriptions.md
pricing-tiers.md
use-cases.md
Creating Your CLAUDE.md Context File
CLAUDE.md tells Claude how to use your HubSpot data.
Key sections:
- Sales process overview — Your pipeline stages and typical sales cycle
- CRM structure guide — Where different types of information live
- Contact context index — Quick reference for active deals
- Template usage guide — Which template to use when
- Sales intelligence — Key patterns Claude should know
Example snippet:
## Sales Process
Average deal cycle: 60-90 days for enterprise, 30-45 days for SMB.
Pipeline stages defined in `/CRM/Pipeline/pipeline-structure.md`.
### Active Opportunities
All active deals tracked in `/CRM/Contacts/active-opportunities.md`. Updated weekly.
High-priority deals this month:
- Acme Corp — $50K enterprise deal, demo scheduled, referred by John
- Beta Inc — $30K SMB deal, proposal sent, price-sensitive
### Email Templates
All templates in `/CRM/Templates/Email/`. Choose based on:
- **Discovery stage** → Use `discovery-outreach.md`
- **Post-demo** → Use `post-demo-followup.md` (enterprise) or `post-demo-simple.md` (SMB)
- **Proposal sent** → Use `proposal-delivery.md`
Response rates tracked in each template file. Stick to proven language.
Workflows That Work
Drafting Follow-Up Emails
Before: You tell Claude "write a follow-up to the demo." Claude writes generic AI slop with no context.
After: Claude reads the contact file, sees it's Acme Corp in Demo Scheduled stage, checks your post-demo template, references their notes about Q1 start date, drafts an email that feels like it came from someone who was actually on the call.
Analyzing Stuck Deals
Before: You manually review each deal, try to remember why it's stalled.
After: You ask Claude "why are three deals stuck in proposal stage?" Claude reads your pipeline definitions, checks the deal notes, identifies the pattern: all three are price-sensitive SMBs who need monthly payment options.
Prepping for Sales Calls
Before: You skim HubSpot right before the call, miss important context.
After: You tell Claude "prep me for the Acme call." Claude reads their contact file, previous interactions, deal stage, surfaces key points: "Referred by John. Pain point is manual reporting. Asked about Q1 start. Budget confirmed at $50K. Watch for procurement process questions."
Training New Reps
Before: New rep shadows calls, reads scattered docs, asks you the same questions repeatedly.
After: New rep asks Claude. Claude reads your playbook and explains qualification criteria, objection handling, closing tactics. New rep learns from your proven process, not generic sales advice.
Maintenance and Updates
Your vault needs regular updates to stay useful.
Weekly: Export updated contact lists for active opportunities. Add notes from significant calls or deal changes.
Monthly: Review win/loss patterns. Update templates if new language is working better. Refresh your pipeline analysis.
Quarterly: Audit your playbook. What's changed? What's working? What needs updating? Revise accordingly.
The time investment is minimal. 20 minutes a week. Your return is AI that actually understands your sales process.
What This Isn't
This isn't a HubSpot API integration. You're not syncing data in real-time or building custom apps.
This isn't a replacement for your CRM. You still work in HubSpot. The vault is Claude's memory, not your system of record.
This isn't automatic. You export manually when data changes significantly. Most CRM context is stable week-to-week.
It's a context bridge. HubSpot holds your data. Claude gets the intelligence to use it.
Give AI Your Sales Context
One markdown file. One afternoon. AI that actually remembers who you are, what you do, and how you work.
Build Your Memory System — $997