SSSahil Sharma
Email
← Back to Thoughts
Wed, 9 Apr, 2025

Unlocking the Power of AI + MCP + Automation

How AI, Model Context Protocol (MCP), and automation tools combine to enable end-to-end workflows using plain English prompts.

From plain English to real automation

In my previous post, I demonstrated how MCP (Model Context Protocol) servers can transform plain English instructions into functional Playwright frameworks.

That was just the beginning.

This time, I explored deeper, end-to-end integrations where AI didn’t just generate code — it executed workflows.

Visual overview

AI, MCP servers, and automation working together in a unified workflow

What I automated using AI + MCP

All of the following tasks were executed using plain English prompts via an LLM, backed by the appropriate MCP servers:

  • Database operations
    Fetched user data directly from MySQL using a database MCP server

  • Data sanitization
    Identified and cleaned duplicate email records through prompt-driven instructions

  • Automated registrations
    Signed up users on a website using Playwright MCP

  • API validation
    Verified registration responses and access tokens via REST API MCP

  • Data logging
    Stored registered emails into an Excel sheet using Excel MCP

  • Version control
    Pushed the complete codebase to GitHub using Git MCP — no manual commands

What made this possible

All of this worked because MCP provides:

  • Structured, auditable tool invocation
  • Secure boundaries between LLMs and real systems
  • Predictable execution instead of free-form automation

Enhanced development workflow

Working inside VS Code Insiders, I was able to:

  • Access all configured MCP servers directly in the editor
  • Generate and modify Playwright scripts in real time
  • Perform Git operations and commits using prompts
  • Seamlessly integrate databases, APIs, and Excel workflows

This is a clear example of how AI-powered agentic workflows can dramatically improve developer productivity.

Organizations embracing MCP

Companies and open-source communities are actively building or adopting MCP servers for real-world AI tasks, including:

  • Atlassian (Jira / Confluence access)
  • Cloudflare
  • Stripe
  • PayPal
  • Intercom
  • Webflow
  • Asana
  • Linear
  • Replit
  • Zed
  • Sourcegraph
  • OpenAI
  • Google DeepMind

MCP is quickly becoming the connective tissue between AI models and real systems.

Explore MCP servers

If you want to explore available MCP servers and real integrations:

  • 🌐 mcpservers.org — curated list of open and community-driven MCP servers
  • 🌐 Live examples and integration guides (linked in my earlier post)

This ecosystem is growing rapidly — and we’re still early.

Final thought

AI alone is powerful.

But when combined with MCP + automation, it becomes something more:

AI that doesn’t just suggest code —
it executes complete workflows, safely and deliberately.