Skip to content
MCP for ChatGPT

Your revenue strategy, inside ChatGPT.

GrowthKit connects to ChatGPT via Remote MCP. Your ICP, pipeline, competitors, and brand voice — permanently stored and used in every conversation. Works with ChatGPT Apps and Custom GPTs.

SIGNAL TO ACTION

Signal

Dreamdata detects intent. A SaaS company visited your pricing page 4 times this week.

GrowthKit Memory

ICP match: 94%. Apollo enriches: VP Sales found. Pipeline history: no prior contact.

Action

ChatGPT drafts personalized outreach using your brand voice and ICP context. Ready to send.

REAL USE CASES

What you can actually do

"Run the weekly review. Show me pipeline changes and new signals."

Weekly Revenue Review

ChatGPT reads pipeline and signals chapters, compares week-over-week, identifies stalled deals, highlights high-fit intent signals.

"Analyze this competitor's website and update our battle card."

Competitor Intelligence

Researches competitor, compares positioning against brand chapter, stores in competitors chapter for future use.

"Create a campaign brief for our Q2 LinkedIn push."

Strategy-Aligned Campaigns

Campaign Brief playbook reads ICP, brand voice, past learnings. Output grounded in defined strategy.

"Find 10 companies in DACH that match our ICP and enrich the top 3."

Targeted Lead Discovery

Uses ICP chapter for search criteria, scores matches, enriches via Apollo (Pro plan).

"Onboard my team. Share strategy access with Head of Marketing."

Team Collaboration

Create Team tokens with permissions. Head of Marketing reads strategy/ICP/brand, writes to campaigns/learnings/analytics.

"What did we learn from the last 3 campaigns? What should we change?"

Accumulated Intelligence

Searches learnings and campaigns chapters including months-old insights. Synthesizes patterns across time.

COMPARISON

ChatGPT Memory vs GrowthKit

ChatGPT Memory (built-in)

  • Stores basic facts and preferences
  • Single user only
  • Flat, unstructured memory
  • No semantic search
  • No version history
  • No CRM or signal integration
  • No guided workflows

GrowthKit Memory

  • 11 purpose-built revenue chapters
  • Team with Admin / Team / View roles
  • Structured, auto-tagged, semantically searchable
  • Semantic search with chapter and tag filtering
  • Full version history with restore
  • Pipedrive, Dreamdata, Apollo integrations
  • 6 guided playbooks that read/write chapters

CHATGPT EXTENSIBILITY OPTIONS

ChatGPT Apps vs Custom GPTs vs MCP

OpenAI gives you three ways to extend ChatGPT for your business. They're often confused, but they solve different problems. Here's how they compare — and why MCP is the right layer for revenue strategy.

ChatGPT Apps

Lightweight in-chat widgets

Embedded mini-apps (Booking.com, Canva, Zillow) that render rich UI inside a ChatGPT conversation. Built on the Apps SDK, which itself uses MCP under the hood.

Best for
Consumer interactions: book a hotel, design a poster, browse listings.
Memory
No persistent business memory. Each conversation starts fresh.
Team
Single user. No roles, no sharing.
Data
App provider's data only. Can't read your CRM or signals.
Setup
  1. Open ChatGPT, type "@" then app name, or pick from app directory
  2. Authorize the app once
  3. Use it inside any conversation

Great for in-chat utilities. Not built for revenue strategy.

Custom GPTs

Static instructions + actions

A configured ChatGPT with a system prompt, knowledge files (up to 20), and optional Actions (REST APIs). Lives in your GPT Store or workspace.

Best for
Repeatable single-user workflows with fixed instructions (e.g., 'My SEO writer GPT').
Memory
Static knowledge files. No structured updates, no version history, no semantic search across chapters.
Team
Per-user creation. Sharing is read-only — collaborators can't write back into the GPT's knowledge.
Data
What you upload + Action endpoints you wire up manually. No native CRM/intent layer.
Setup
  1. ChatGPT → Explore GPTs → Create a GPT
  2. Write instructions, upload knowledge files
  3. (Optional) Add Actions via OpenAPI schema for each external API
  4. Publish to workspace or GPT Store

Useful for personal assistants with fixed scope. Breaks down for evolving team strategy.

MCP (Remote Connector)

Recommended

Live, structured, team-shared memory

Model Context Protocol — an open standard (Anthropic-originated, adopted by OpenAI) that exposes tools and resources to ChatGPT via OAuth. Works in regular chats, Custom GPTs, and Apps.

Best for
Team-wide revenue intelligence: ICP, pipeline, brand voice, signals — read and written by every conversation.
Memory
Persistent, structured chapters. Semantic search, version history, auto-tagging.
Team
Admin / Team / View roles. One source of truth across the org.
Data
Native integrations with Pipedrive, Dreamdata, Apollo. Tool calls retrieve live data.
Setup
  1. Settings → Connectors → Add custom connector
  2. Paste mcp.growthkit.tools, complete OAuth (Dynamic Client Registration)
  3. ChatGPT gains 20+ tools instantly — no per-API wiring
  4. Run the onboarding playbook once; reuse forever

The right layer for a Revenue Operating System.

WORKED EXAMPLES

Two real workflows, end to end

Below are two concrete examples we run with B2B revenue teams in DACH. Each shows the trigger, the prompt, what ChatGPT does via MCP, and the artifact it produces.

Example 1 — Inbound signal triage

Turn a Dreamdata intent spike into a personalized outreach email

On Monday morning, a German SaaS company (180 employees, HR-tech) has visited your pricing page 4 times and your integrations page twice in 7 days. Dreamdata flags it as a high-intent account. You have ~3 minutes before standup.

  1. 1

    Prompt in ChatGPT

    "Pull this week's top intent signals from Dreamdata, match against ICP, and draft outreach for the #1 account in our brand voice."

  2. 2

    What MCP does

    1) Calls Dreamdata via MCP tool to fetch top 10 signals. 2) Reads ICP chapter (target: B2B SaaS, 100–500 FTE, DACH, HR/People-tech). 3) Scores account = 94% match. 4) Reads Brand Voice chapter (tone: direct, outcome-led, no buzzwords). 5) Reads Pipeline chapter to confirm no prior contact.

  3. 3

    Apollo enrichment

    Identifies VP People Ops as primary persona; finds verified email and 2 mutual LinkedIn connections.

  4. 4

    Output you get

    A 90-word German email referencing the prospect's recent funding round, your HR-tech ICP fit, and a specific use case from your Learnings chapter where a similar account closed in 11 days. Copy-paste ready.

  5. 5

    Time saved

    ~25 min of manual research + writing → ~90 seconds. The strategic context (ICP, brand, learnings) is reused, never re-briefed.

Example 2 — Weekly revenue review

Run a Monday revenue review across pipeline, signals, and learnings

Every Monday at 9:00, the revenue team needs: pipeline movement vs last week, top 3 stalled deals, top 3 new high-fit signals, and one experiment proposal grounded in past learnings. Previously a 60-minute manual exercise across Pipedrive, Dreamdata, and a Notion doc.

  1. 1

    Prompt in ChatGPT

    "Run the Weekly Revenue Review playbook for the last 7 days."

  2. 2

    What MCP does

    1) Reads Pipeline chapter + live Pipedrive data via tool call → diff vs prior week. 2) Reads Signals chapter + Dreamdata feed → top 3 new accounts above ICP threshold. 3) Reads Learnings chapter → finds 2 patterns relevant to current stalled deals. 4) Writes the review back into the Analytics chapter (versioned).

  3. 3

    Output you get

    A structured brief: 5 pipeline deltas with €-impact, 3 stalled deals with diagnostic + suggested next step, 3 new signal accounts with persona + ICP-match score, 1 experiment proposal citing the specific past learning it builds on.

  4. 4

    Team handoff

    Head of Marketing (Team role) reads the Analytics chapter, picks up the experiment proposal, and writes the campaign brief — no re-briefing of strategy needed.

  5. 5

    Time saved

    ~55 min/week → ~3 min. More importantly: every review compounds, because outputs are stored as new context for next week.

GET STARTED

Connected in 5 steps

1

Create your GrowthKit account

Sign up, free Starter with 50 memories and 5 chapters.

growthkit.tools/signup
2

Add MCP in ChatGPT Settings

Settings → MCP/Connectors → Add Remote MCP.

mcp.growthkit.tools
3

Authenticate via OAuth

Standard OAuth with Dynamic Client Registration. Approve access, ChatGPT gains 20+ tools.

4

Run the onboarding playbook

Type "Run the GrowthKit onboarding playbook". Describe business, ICP, brand voice. Stored in right chapters.

"Run the onboarding playbook"
5

Experience the compound effect

Next week, ask for weekly review. Watch ChatGPT pull from pipeline, signals, strategy without reminders.

Frequently Asked Questions

Via Remote MCP with OAuth. Settings → Add MCP → mcp.growthkit.tools → OAuth flow. Under 2 minutes.

Stop briefing ChatGPT.

Start working with it.

Connect GrowthKit and give ChatGPT the revenue context it's been missing.

Also available for Claude