Using MCP with Plecto
Plecto connects to AI assistants through the Model Context Protocol (MCP), an open standard that lets AI tools read and manage your Plecto data through natural conversation. Once connected, you can create dashboards, build formulas, and add widgets by describing what you want instead of writing formula syntax manually.
Getting connected
You’ll need MCP enabled for your organization before you can connect an AI assistant.
The Plecto MCP endpoint is https://app.plecto.com/api/mcp.
Connect from claude.ai
Open Settings → Connectors → Add custom connector.
Enter https://app.plecto.com/api/mcp.
Sign in when redirected and choose the Plecto organization you want to connect.
Click Allow to grant access.
On first use, Claude Desktop opens a browser window so you can complete login and authorization.
Connect from other MCP-compatible clients
Any AI client (like ChatGPT) that supports remote MCP servers over HTTP with OAuth can use the same endpoint: https://app.plecto.com/api/mcp. Follow your client’s instructions for adding a remote MCP server. During authorization, you’ll be redirected to Plecto to sign in and select an organization.
What you can do
Dashboards
Browse all dashboards or search by name
Create a new dashboard and choose a theme
Start from a suggested layout, such as a balanced layout with KPI tiles, charts, and detail tables
Update dashboard settings including title, theme, teams, and time zone
Delete dashboards you no longer need
Example prompts
“Show me all my dashboards”
“Create a Sales Performance dashboard and add a number widget showing total revenue this month”
“Delete the test dashboard I made last week”
Widgets
Your assistant can add and manage multiple Plecto widget types:
Number tiles for a single KPI, optionally with red, yellow, and green thresholds
Speedometers for progress toward a target
Graphs including area, line, column, pie, donut, and funnel charts
Leaderboards ranking employees or teams by a metric
Tables and timelines with multiple columns backed by formulas
Text widgets for notes, headings, and instructions
Timers for deadlines or live clocks
Streaming widgets for live external values
Charts can be grouped by employee, team, time such as day, week, or month, or by custom fields from your data source.
The assistant checks the existing layout before adding anything, so new widgets do not overlap existing ones. You can also ask it to resize, move, or replace widgets.
Example prompts
“Add a leaderboard to my Sales dashboard showing top reps by closed deals”
“Make the revenue tile turn red below 50,000 and green above 100,000”
“Add a donut chart of calls split by direction”
“Add a column chart of won deals per month for this year”
“Build a funnel widget showing leads, qualified, proposal, and won stages from my CRM”
“Add a countdown timer to the end of the quarter”
“Move the leaderboard to the bottom row and make it span two columns”
“Replace the old revenue widget with a speedometer”
Formulas
Browse existing formulas, including folder organization
Create new formulas from plain-language descriptions
Choose display formats such as numbers, percentages, durations, dates, prefixes, or suffixes
Check a formula’s value for current or past periods
Break results down by employee or team
Test a formula without saving it
Edit or delete formulas
Automatically organize dashboard-related formulas into a folder named after the dashboard
Example prompts
“Create a formula for average deal size on won deals in Pipedrive, shown in dollars”
“What’s the current value of my Monthly Revenue formula, and what was it last month?”
“What’s my Win Rate per team this quarter?”
“Test this for me without saving: count of open Zendesk tickets where priority is high”
Data sources, teams, and themes
List connected data sources and inspect their fields, types, and allowed values
View organization teams for dashboard sharing or grouping
Browse available dashboard themes, including custom themes
Example prompts
“What data sources do I have connected?”
“What themes are available? Switch this dashboard to the dark one”
How it looks in practice
Build a dashboard from one prompt
A single prompt can be enough to create a complete dashboard. For example, asking “Can you make me a dashboard for Aircall Calls in Plecto?” can lead the assistant to explore the data source, create formulas, and place widgets automatically.
The resulting dashboard can include KPI tiles at the top, charts in the middle, and a leaderboard below.
Edit the layout in plain language
Follow-up instructions work the same way. If you ask the assistant to “replace the middle row with a funnel chart”, it can remove the existing widgets in that row and place a new full-width funnel widget in the same space.
Ask questions about your data
You can also use the assistant for ad hoc analysis without creating widgets or saving formulas. Questions like “Who made the most calls in February?” can be evaluated directly against your data.
Explore KPIs automatically
You can ask the assistant to look for useful insights in a data source, evaluate promising KPIs, and suggest widgets worth adding to a dashboard.
Tips for better results
Name the data source clearly if you have multiple integrations. “Calls from Aircall” is more precise than “calls.”
Provide a dashboard name if you already know it. Otherwise, the assistant can suggest one.
Reference existing formulas by name when building on previous work.
Ask it to test first if you want to validate a formula before saving it.