Hands-On MCP Workshop: MCP as a GTM strategy
Bring your product, API, or idea. Learn how to turn one useful capability into an MCP tool that AI assistants can understand and use.
Hosted by Dean Grover, Founder at chanl.ai with CAST-GNY and Bond AI.
Most people meet MCP as a developer feature: a way to let Claude use your tools. That is true, but it undersells it.
Build one MCP server and it can connect your product to Claude, ChatGPT, Cursor, Copilot, Gemini, Perplexity, and many other AI clients.
One protocol. One build. Integrated with many AI assistants with millions of users.
That makes MCP more than an integration. It could become a new distribution channel for your product.
The honest version is that being connectable everywhere does not make you discoverable everywhere. Each platform has its own directories, approval processes, and rules.
Both sides are worth understanding before you build.
Your tool description is your landing page - and the AI agent or the model is the buyer reading it.
That is why this workshop is both technical and commercial. We will look at what to expose, how to build it, how models decide to use it, and what changes when you make it available to others.
The hands-on challenge
You will work individually or in a small group to turn one product capability into an MCP tool.
You will:
Choose one useful action from a product or API
Define when an assistant should and should not use it
Write the tool name, description, inputs, and response
Build and test it locally
See what happens when the model gets it wrong
Explore different ways to publish it
Add a security or safety guard
Bring your own product or use one of the examples provided.
Connect, Build, and Publish
Connect
Start by using an existing MCP service through a platform such as Composio.
You will connect an application, make its tools available to an assistant, and see how the model decides what to call.
Build
Create your own MCP tool locally using FastMCP.
The goal is not to build a complete integration. It is to create one useful tool and test whether the model understands:
When to call it
What information it needs
When not to call it
What to do when something fails
Publish
Learn how a local MCP server becomes a remote service that other people and assistants can access.
We will compare infrastructure options such as Cloudflare and Vercel with managed alternatives.
Protobox.ai, a beta product, will be available as one optional managed path for publishing and testing your MCP tools - with a free pro account for attendees.
What we’ll learn
How MCP connects AI assistants with products and services
How to choose the right product capability to expose
How models decide when to call a tool
Why tool names, descriptions, and schemas matter
How to build locally with FastMCP
How remote MCP deployment works
How Cloudflare, Vercel, and managed platforms differ
How to test ambiguous, incorrect, and unsafe requests
Where MCP fits into product strategy and go-to-market
Security and production Readiness
A local demo and a customer-facing MCP server are not the same thing.
We will discuss:
Authentication and user permissions
Protecting credentials and API keys
Input validation
Read-only versus write access
Confirmation for sensitive actions
Logging, monitoring, and rate limits
You will also test whether another participant can make your tool perform an action it should reject.
Who It’s For
Founders exploring MCP as a product or distribution channel
Product managers deciding what assistants should be able to do
Engineers looking for practical MCP experience
GTM, growth, and developer relations teams evaluating MCP
Anyone who has experimented with MCP but has not published a tool
No prior MCP experience is required.
Non-technical attendees can focus on the use case, tool design, testing, and product strategy while pairing with a builder.
What You’ll Leave With
Experience using an existing MCP service
A clearly defined MCP use case
A local MCP server built with FastMCP
Practical experience testing model tool calls
An understanding of different publishing options
A security and production-readiness checklist
Optional access to a free Protobox beta account
Agenda
5:30–6:00 PM
Arrival, setup, and pairing
6:00–7:45 PM
Presentation and Hands-on workshop
7:45–8:15 PM
Demos and wrap-up
Before You Arrive
Bring:
A laptop and charger
A Claude account
One idea for something an AI assistant should do with a product
If you plan to code:
Python 3.10 or later
A code editor
A GitHub account
Example use cases and starter projects will be available.
Workshop Lead
Dean Grover is founder at Chanl.ai.
He has spent more than two decades building software products and platforms, including systems delivered through Cisco and Paramount that reached more than 100 million homes.
Today, he builds AI products and infrastructure full-time. He also runs an AI Incubator for Solopreneurs @ Fabrik NYC.
Join us and see all upcoming events at luma.com/chanl
Connect with Dean and ask questions about the event or any help on these topics at https://www.linkedin.com/in/deangrover/
Logistics
Free to attend. Capacity is limited because this is a hands-on workshop.
Fordham Gabelli School of Business
140 West 62nd Street
Tiered Lecture Hall 333
Bring a photo ID and allow time for building security.
Food and drinks are not permitted inside the lecture hall.
If your plans change, please release your registration so someone on the waitlist can attend.
CAST-GNY
Chinese Association for Science and Technology USA(CAST-USA) was founded in 1992 in New York, as an US nonprofit, nonpolitical organization. CAST-USA Greater New York(CAST-GNY) is the founding and one of the most active chapters among CAST-USA’s 16 chapters, with a network of close to 10000. It aims to promote learning, innovation, and exchange/cooperation among science and technology professionals and entrepreneurs. It’s a platform for education, networking and mentoring, focus on career development/entrepreneurship and empowering next generation in Science & Technology and leadership, through events such as knowledge sharing seminars, career development events, entrepreneurship exchange, youth STEM project competition and summer camps, startup competition and annual summit.