Model Context Protocol (MCP)

Estimated reading: 5 minutes 171 views

The Model Context Protocol (MCP) is an open-source standard that allows AI models to seamlessly connect with external data, tools, and software systems. It acts as a universal bridge, enabling LLMs to securely access files, databases, and APIs, fostering the development of AI agents that can take action rather than just generate text.

Modal Context Protocol (MCP) enhances Ask Me by transforming it from a limited Q&A interface into a unified conversational AI assistant. With MCP, Ask Me can access external market intelligence, automatically understand user intent, and dynamically determine the most appropriate response format, whether narrative insights, visual charts, or a combination of both. It can also intelligently orchestrate internal Ask Me services, enabling more flexible, context-driven analysis, such as ChatGPT-style root cause analysis (RCA). By automatically handling agent selection and response modes, MCP delivers a seamless, single AI experience where users no longer need to manually switch between agents or analytical modes.

With MCP enabled, Ask Me overcomes the following limitations:

Limited access to external insights: Previously, Ask Me could not fetch external market trends or global intelligence (e.g., trends impacting agri-commodities in India). MCP enables integration with external intelligence sources to enrich insights.
Restricted output formats: Earlier responses were constrained to predefined cards and flows. MCP allows more flexible outputs, including narrative insights, visual charts, or a combination of both.
Manual agent switching: Users previously had to switch between agents, such as NLQ and Data Science, depending on the request. MCP automatically detects intent and selects the appropriate capability without user intervention.
Rigid RCA processes: Root Cause Analysis followed a fixed step-based workflow. MCP enables more flexible, conversational RCA that supports contextual drill-down and natural exploration.
Inability to orchestrate services: Earlier, Ask Me could not coordinate actions across platform services or tools. MCP introduces the ability to call and orchestrate internal services within the conversation.
Passive assistant experience: Ask Me previously behaved like a basic Q&A interface. MCP transforms it into a more intelligent assistant capable of reasoning, adapting responses, and supporting decision-making workflows.

To enable and use MCP capabilities:

Step 1: Enabling Extended Capabilities switches Ask Me to MCP mode.

Step 2: Now choose the required schema for your analysis.

Click “Preview Data” to view the schema data.

Note:

Schema info: The Schema Info section provides an overview of the data structure available within the selected schema. It helps users understand what data elements can be used while asking questions in Ask Me. It includes attributes, location, measures, etc.

Query Guidelines: Query Guidelines provide recommendations on how to frame questions so that Ask Me can understand and process them more accurately. These guidelines help users write clear, structured queries using the available schema fields, ensuring better insights and more relevant results.

Step 3: You can ask your question by typing it in the input box or using the audio feature.

How much Operating Profit is generated by each Division?

Which property and division are most profitable for our business

Core Capabilities

Market Intelligence Integration
MCP enables Ask Me to access and retrieve external intelligence, including:

- Market trends
- Industry updates
- Global insights
- Competitive intelligence

Example: “What are the current market trends impacting the Digital division?”

With MCP enabled, Ask Me can respond by combining external market intelligence with relevant internal data context, providing more comprehensive and informed insights.

Automatic Mode Detection (No Manual Switching)

MCP automatically detects the most suitable response format based on the user’s query, such as Narrative mode, Visual mode, or a Hybrid format (Text + Chart). This eliminates the need for users to manually switch to the NLQ agent, change narrative modes, or select chart options. MCP intelligently determines the appropriate mode and delivers the response accordingly.

Example A: “Why did promo costs increase in the Las Vegas Strip region in January 2024?”

Example B: “Analyze promo cost trends across properties in the Las Vegas Strip.”

Flexible Output Formats

With MCP enabled, responses can be presented in multiple formats depending on the user’s query. These may include text-only insights, chart-only visualizations, a combination of text and charts (Text → Chart → Text), multiple chart cards, or narrative-driven Root Cause Analysis (RCA) for deeper explanation and context.

All charts continue to support the existing card functionalities, including options to download, expand, modify settings, and use other standard card features, ensuring a consistent and interactive analysis experience.

Example: “Compare promo cost by region and property for January 2024.”

Flexible Root Cause Analysis (ChatGPT-Style)

Unlike traditional RCA processes that follow rigid, predefined steps, MCP enables a more flexible and conversational approach to analysis. It supports context-aware reasoning, free-flow explanations, structured narratives, and logical drill-downs based on the data rather than a fixed analytical flow.

For example, when a user asks, “Why were sales low in Q4 2024 in Brazil?”, MCP can analyze the context and provide a comprehensive explanation by examining multiple factors, including demand-side changes, product mix variations, seasonal trends, the impact of discount strategies, and historical performance patterns. This approach delivers insights in a natural, conversational format, making the analysis easier to understand and more aligned with how users explore data.

Example: “Why is promo cost higher for Caesars Palace compared to other properties?”

Unified Ask Me Service Calling (Internal Tool Orchestration)
- Automatically utilizes available Ask Me services to process queries efficiently.
- Prevents unnecessary triggering of the Data Science (DS) agent, ensuring smoother query handling.
- Invokes the NLQ engine intelligently when required, based on the user’s request context.

Example:
Previously, queries containing terms like “Change” or “Trend” would automatically trigger the DS agent. With MCP enabled, the system interprets the query context and automatically executes the most appropriate logic, without requiring manual agent switching.