Episode 81 · Season 1
MCP Apps: Your AI Assistant Can Finally Show You a Screen, Not a CSV
Ask an AI assistant for every sale in your region and you get a CSV, or a table the model improvised, and then you scroll and Control-F through it like a spreadsheet. Tim Harting and Gavin Franklin of Cirrius Solutions break down MCP Apps, the extension that lets an MCP server send back a working interface along with the data, and show where that changes the daily work of the people using these tools.
What you’ll learn:
- How an MCP App differs from a plain MCP call: the server returns an HTML interface with filtering, sorting, search, and buttons that trigger actions, instead of raw data the model has to format on its own
- What is happening under the hood: the interface talks to the model and back to the MCP server, and the whole exchange stays inside that same locked-down loop
- Where early use is landing: data exploration, scenario and sales-forecast modeling, media and diagram rendering, plus Gavin’s read on token cost
Who it’s for:
Technology and engineering leaders who already run MCP connections and want to know what this extension changes for their users, and product and business leaders deciding which workflows belong inside a chat interface.
Guest + credibility:
Tim Harting is COO of Cirrius Solutions and Gavin Franklin is its Head of Custom Development. Cirrius is a Salesforce shop that works through Salesforce’s hosted MCP service, and Gavin’s team is already implementing scenario-modeling and forecasting interfaces on MCP Apps.
Primary CTA:
Listen, then open the companion artifact that goes with this episode; it lays the whole concept out in bite-sized pieces. As Gavin puts it, with the right interface you don’t even necessarily need to know what to ask.
Recommended Links & Resources:
Tim Harting on LinkedIn
Gavin Franklin on LinkedIn
[Add: episode page URL (companion artifact)]
Greg Banks on LinkedIn
Jason Fowler Music
Cirrius Solutions
Agentic AI Episodes
Any Questions: Reach out to us at CirriusTalk@cirriussolutions.com
We cover
- Understanding MCP Apps and Their Functionality
- Use Cases and Examples of MCP Apps
- Security and Token Usage in MCP Apps
- Future of MCP Apps and Their Impact
- Understanding Token Spend and User Interaction
Chapters
- 0:00Introduction to MCP Apps and Agentic AI
- 0:36Introduction to MCP Apps
- 2:07Understanding MCP Apps and Their Functionality
- 5:12Use Cases and Examples of MCP Apps
- 9:11Security and Token Usage in MCP Apps
- 12:36Future of MCP Apps and Their Impact
- 16:53Conclusion and Reflections on AI Evolution
- 17:47Introduction to Cirrius Talk and GenTic AI
- 18:16Understanding Token Spend and User Interaction
Resources and downloads
The artifacts Greg’s team built for this episode, free to download.
Beyond Tool Calls: Inside MCP Apps
Explains MCP Apps, the official Model Context Protocol extension that lets a tool call return an interactive interface instead of plain text or JSON. Shows how one request replaces four rounds of re-prompting, how the interface runs in a sandboxed frame inside hosts like Claude, VS Code Copilot and Microsoft 365 Copilot, and where it is already used: data exploration, scenario and budget modeling, media viewers and live diagramming.
Related episodes
Agentic AIJev vs LLMs: Decisions in Milliseconds for 4 Cents per Million Tokens
Agentic AIThe Artifact Ladder: 5 Tiers From PowerPoint to Full AI-Built Apps
Agentic AIClaude vs GPT-5.6: How to Choose the Model Behind Your AI Agents
The Military MinuteFrom Homeless to Homeowner: Inside a Mighty Hero Homes BaseDrew Walston, Founder
Salesforce at ScaleLisa Malmanger – From Army Sergeant to Salesforce Solution ArchitectLisa MalmangerTalk to us about AI agents
Keep going: See how we build AI agents for real businesses, Talk to us about a custom AI agent, More from the Cirrius blog