πŸ”Œ MCP

Connect everything
to intelligence.

The Model Context Protocol (MCP) allows AI to interact directly with your tools, APIs and services. More than a chatbot β€” an assistant that reads, writes and acts in your systems. A tool designed for professionals.

πŸ’¬ AI Chat β€” MCP active

GitHub Slack Notion

What are the open issues on the frontend repo and create a summary on Notion

πŸ“‘ github.list_issues πŸ“ notion.create_page

I found 12 open issues on frontend-app:
β€’ 4 bugs (2 critical)
β€’ 5 feature requests
β€’ 3 improvements

βœ… Summary created on Notion β†’ Frontend Issues β€” February 2026

Send the summary to #dev-team on Slack

πŸ’¬ slack.send_message

βœ… Message sent to #dev-team with the recap of 12 issues.

What is MCP for?

AI moves from words
to action

The Model Context Protocol (MCP) is an open standard that allows AI models to connect directly to your external tools and services. Without MCP, AI can only answer your questions. With MCP, it can read your files, query your databases, create Jira tickets, send Slack messages, update your Notion documents β€” all from a simple conversation in Oupi Chat.

It's a tool designed for professionals who want to automate complex tasks without leaving their AI interface.

πŸ”Œ With and without MCP

❌ Without MCP

β†’ "What are my GitHub issues?"

"I don't have access to GitHub. You can check github.com/..."

AI cannot act

βœ“ With MCP

β†’ "What are my GitHub issues?"

"You have 12 open issues: 4 bugs, 5 features, 3 improvements..."

AI reads, acts and reports

βš™οΈ MCP Architecture on Oupi

πŸ‘€ You β€” Oupi Chat
🧠 Oupi AI + Auto Model Selector
πŸ”Œ MCP Layer β€” Secure protocol

GitHub

Repos, Issues, PRs

Slack

Messages, Channels

Notion

Pages, Databases

Jira

Tickets, Sprints

Google Drive

Docs, Sheets

Your APIs

Custom MCP

How it works on Oupi

How MCP works
on Oupi

On Oupi, MCP works in 3 steps. First, you connect your MCP servers (GitHub, Slack, Notion, Jira or your own APIs). Then, when you ask a question in Chat, AI automatically detects if it needs an external tool to respond. Finally, it calls the corresponding MCP server, executes the action and reports the result β€” all in a natural conversation.

Connect your MCP servers in a few clicks
DΓ©tection automatique des outils nΓ©cessaires par l'IA
Real-time action execution in your systems
Results naturally integrated into the conversation
Chaining possible: multiple tools in a single request
MCP Servers

Dozens of tools
connectable

Connect the tools your team already uses. Each MCP server gives AI specific capabilities.

GitHub Development

Read/create issues, PRs, repos. Search code. Manage branches and deployments.

Slack Communication

Envoyer des messages, lire des canaux, rechercher dans l'historique, crΓ©er des fils de discussion.

Notion Documentation

Create and edit pages, query databases, organize projects.

Jira Project management

Create tickets, update sprints, read backlogs, generate reports.

Google Drive Files

Read, create and edit documents, spreadsheets, presentations. File search.

HubSpot CRM

View and update contacts, deals, sales pipelines.

Salesforce CRM

Access accounts, opportunities, reports. Update CRM data.

Databases Data

Query MySQL, PostgreSQL, MongoDB in natural language. Secure read/write.

Your APIs custom Custom

CrΓ©ez votre propre serveur MCP pour connecter n'importe quel outil ou API interne Γ  l'IA.

MCP + AI Chat

Chat becomes
your command center

With MCP enabled in Oupi Chat, you no longer need to open 10 tabs. Ask AI to check your GitHub issues, create a summary on Notion, post a message on Slack and update a Jira ticket β€” all in a single conversation. AI understands context and chains actions automatically.

Direct access to your tools from Chat
Automatic chaining of multiple actions
Shared context between connected tools
L'IA mΓ©morise vos prΓ©fΓ©rences et habitudes
Compatible avec tous les modèles IA via l'Auto Model Selector
Discover AI Chat β†’

⚑ MCP action examples

"Create a bug ticket on Jira for the login bug"

🎫 jira.create_issue β†’ Ticket BUG-347 created

"What are the latest messages from #general?"

πŸ’¬ slack.read_channel β†’ 8 recent messages

"Add this report to our Notion wiki"

πŸ“ notion.create_page β†’ Page created in Wiki

"How many open deals in HubSpot?"

πŸ“Š hubspot.list_deals β†’ 23 deals for €840K

"Deploy the main branch to staging"

πŸš€ github.create_deployment β†’ Deployed v2.4.1

πŸ” Permissions & Control

GitHub Server

Read repos βœ“ Create issues βœ“ Merge PRs βœ— Delete repos βœ—

Slack Server

Read channels βœ“ Send messages βœ“ Private channels βœ— Delete msgs βœ—
πŸ”’ Each user has their own permissions
Security

Full control
over every action

You precisely define what AI can and cannot do in each tool. Permissions are granular per user, per server and per action. Sensitive actions (deletion, merge, private channel posting) can be blocked or require confirmation.

Read/write permissions per tool
Sensitive actions blocked or require confirmation
Journal d'audit de toutes les actions MCP
Tokens d'accès chiffrés et renouvelés automatiquement
Compatible with Privacy Protection: data tokenized even via MCP
SDK & Documentation

Create your own MCP servers

Have internal tools? Create a custom MCP server in a few hours with our SDK. AI will then be able to interact with any system in your company.

🐍

Python SDK

pip install oupi-mcp β€” Complete SDK with examples and interactive documentation.

πŸ“¦

Node.js SDK

npm install @oupi/mcp β€” Native integration with your JavaScript and TypeScript APIs.

🐘

PHP SDK

composer require oupi/mcp β€” Perfect for existing Laravel and PHP applications.

MCP is connected to the entire Oupi ecosystem

AI connected
to your ecosystem.

The Model Context Protocol opens AI to all your existing tools. A professional tool for concrete results.

Try for free Contact us