Segments
Describe any audience in a sentence.
“Members who visited 3+ times last month but haven’t booked since.”
Live segment, counting itself — 1,204 contacts
New · Agentic MCP
Run Your Entire Business in Plain English
Connect your account to Claude, ChatGPT, Gemini — or any AI tool that speaks MCP. No dashboards. No API keys. No training. Just ask.
Works with
Claude
ChatGPT
Gemini
Copilot
Cursor
+ any MCP client
Duplicated campaign “Spring Promo — Week 2”
Built segment Non-openers — 2,418 contacts
Scheduled send — Tue, 10:00 AM local time
Done — awaiting your approval to send.
Before & after
Seven Steps, or One Sentence
Every operator has said this out loud. What happens next is the whole difference.
Before Agentic MCP
“I need a campaign for members who haven’t visited in 30 days.”
Then someone has to…- Build the segment
- Create the email
- Create the SMS
- Build the flow
- Configure the delays
- Test it
- Schedule it
Seven screens, and one person who knows where everything lives.
After Agentic MCP
“Build a 30-day win-back campaign for members who haven’t visited. Use email and SMS, offer a free guest pass, exclude anyone already contacted this month, and send me a test.”
- Segment built — already excluding this month’s contacts
- Email and SMS written, with the guest-pass offer
- Multi-step flow assembled and timed
- Test sent to you for approval
Done. One sentence.
Describe it → it exists
It Doesn’t Just Answer. It Builds.
The things that used to mean a builder, a wizard and a dozen decisions get made from a description. No dragging boxes. No blank canvas.
Flows
Multi-step journeys built by describing them, not by dragging boxes.
“A 5-step onboarding journey for new members, across their first 30 days.”
Journey assembled, delays timed, live
Landing Pages
Describe the page, get it built and published.
“A signup page for the summer challenge, with a form and the offer up top.”
Page built, form wired, published
A day’s work in four prompts
What Used to Take an Afternoon
These are not demos. They are the same four requests operators send their teams every week — typed once, executed in seconds.
“Who churned last month — and what did we try to save them?”
Retention review, done.
“Nurture flow for 60-day cold leads — Downtown only.”
Live in seconds.
“Reassign Sarah’s overdue tasks to Mike.”
Team unblocked.
“Draft next week’s lapsed-member campaign. Send me a test.”
Ready for review.
The full surface area
Everything It Can Reach
The Agentic MCP Server exposes your entire Gleantap workspace to your AI — every object, every action, every report. If your team can do it in the dashboard, your AI can do it by prompt.
Contacts
Look up, add, merge, tag — with the full activity timeline behind every profile.
Segments
Describe any audience in a sentence. Get a live, refreshing count back instantly.
Messaging
Send SMS, email, or voice — to one person or a list of fifty thousand.
Campaigns
Create, schedule, duplicate — then pull opens, CTR, and revenue back out.
Templates
Build, preview, and test-send. Merge tags included and validated.
Flows
Multi-step journeys built by describing them, not by dragging boxes.
CRM & Sales
Move deals, assign tasks, and automate sales rules across every pipeline.
Tasks
Create, reassign, and close out tasks — one-off or recurring, for anyone on the team.
Inbox
Read, reply, assign, and close conversations across every channel you run.
Forms
Publish advanced forms and capture new leads without touching a builder.
Landing Pages
Describe the page, get it built and published — then point a campaign straight at it.
Appointments
Book, reschedule, and cancel — fully calendar-aware and conflict-safe.
Events & Webhooks
Log check-ins and purchases. Inspect and replay webhook deliveries.
Scorecards
Build scorecards and pull the numbers back out — no report builder required.
Multi-Location
Each user sees only their own locations. Always. No exceptions, no leaks.
Not a chatbot
A True Agentic Layer
Most “AI in your CRM” is a chat window that summarizes what it can see. This one signs in as you and does the work.
Acts as the user
Real permissions, real workroom scope. It can never reach data you couldn’t open yourself.
Writes real data
Validated on the way in, and never double-fired. Idempotent by design.
Builds things
Segments, templates, flows, forms — created from a description, not a wizard.
You approve
Destructive actions always ask first. Nothing irreversible happens without a yes.
Everything a human does in the dashboard, your AI can do — faster.
5+ hrsSaved per user, weekly*
10xFaster on common workflows*
6 wksWon back per user, yearly*
Day 1Productive — zero training
*Estimates based on typical task times for a mid-size operator (segments, campaigns, reporting, flows, CRM upkeep) vs. the same actions by prompt. Gartner reports teams use only ~33% of their marketing platform’s capabilities — plain-English access removes the learning curve that keeps the rest locked.
Live in minutes
Three Steps. No Engineering Ticket.
No API keys to mint, no middleware to host, no developer to borrow.
1
Connect
Add Gleantap to Claude, ChatGPT, Gemini — or whichever MCP client your team already uses.
2
Authorize
Sign in once. The AI inherits your exact role, permissions, and locations.
3
Ask
Type it in plain English. Review what it plans to do. Approve. Done.
One connection, every assistant. Because Gleantap speaks the Model Context Protocol — the open standard for connecting AI to real systems — it works with Claude, ChatGPT, Gemini, Copilot, Cursor and every other MCP-compatible tool, and with whatever adopts MCP next. No rebuild on your side. Read the developer docs →
Governed by default
Your Guardrails Come With It
The MCP server does not sit beside your permission model — it runs inside it.
Scoped to the signed-in user
The AI cannot see a record, list, or location that the person behind it cannot already see.
Approval before damage
Deletes, bulk edits, and live sends stop and ask. Read operations run freely.
Full audit trail
Every prompt-driven action is logged and attributed, exactly like a dashboard action.
No key sprawl
OAuth, not API keys pasted into config files. Revoke access in one click, any time.
Questions
The Things Operators Ask First
What exactly is an MCP server?
MCP — the Model Context Protocol — is an open standard for connecting AI assistants to the systems where your work actually lives. A Gleantap MCP server means your AI can read and act on your Gleantap workspace directly, instead of you copying data back and forth.
Do I need a developer to set this up?
No. You connect it from your AI client and sign in with your normal Gleantap credentials. There is nothing to host, no keys to generate, and no code to write.
Can the AI delete something by accident?
Destructive actions always stop and ask for explicit confirmation before they run. Reads and drafts happen freely; anything irreversible waits for your approval.
Does it respect multi-location permissions?
Yes, strictly. The AI operates with the exact role, workroom, and location scope of the person who authorized it. A franchise manager's AI sees that manager's locations and nothing else.
Which AI assistants does it work with?
Any assistant that supports the Model Context Protocol — including Claude, ChatGPT, Gemini, Microsoft Copilot and developer tools like Cursor. You are not picking a vendor: connect Gleantap once, and use whichever assistant your team prefers. As more tools adopt MCP, they work too, with no change on your side.
Is this included in my plan?
The Agentic MCP Server is a $49 per month add-on, available on every Gleantap plan. Talk to us about enabling it for your workspace and we will walk through what it unlocks for your team.
See It On Your Data
Book a demo. Run your first prompt today.