Claude MCP Integration: conversational AI to enhance tactical analysis with Kognia

Football clubs have never had access to more data than they do today. Every match generates thousands of events, tracking points, tactical actions and performance metrics, offering unprecedented opportunities to understand players, teams and playing styles.

Yet, despite the abundance of information, accessing meaningful insights often remains a slow and technical process. Coaches, scouts and sporting directors frequently rely on analysts to answer ad-hoc questions, with teams often limited by predefined dashboards and reports. Valuable insights exist, but they are not always easy to reach.

Kognia’s Claude MCP Integration changes that experience entirely.

By combining Kognia's Semantic Layer with Model Context Protocol (MCP), football professionals can interact with their Kognia dataset using natural language. Instead of navigating dashboards or writing SQL queries, they simply ask a question and receive an accurate, contextual answer in seconds.

Rather than replacing existing analytical tools, the integration extends them, making football intelligence more accessible, more flexible and available through conversation.

The Foundation: Kognia's Semantic Layer

Large Language Models are exceptionally good at understanding natural language, but they do not inherently understand how a football database is organised. They cannot know which tables contain tactical events, how a KPI is calculated, or which metrics should always be interpreted together.

This is where Kognia's Semantic Layer becomes essential.

Sitting between the raw database and Claude, the Semantic Layer transforms technical data structures into meaningful football concepts. Instead of exposing database tables, columns and relationships, it provides a curated layer of trusted football definitions such as unmarking movements behind the defensive line, high-intensity actions, pressure on the ball possessor, defensive recoveries and player performance KPIs. 

More importantly, every metric has a single definition.

Whether data is accessed through SQL, Tableau, the Kognia web application or Claude itself, the underlying calculation remains exactly the same. The Semantic Layer acts as the organisation's single source of truth, ensuring consistency, reliability and confidence in every answer.

This allows Claude to reason about football rather than databases.

Why MCP?

The MCP provides the bridge between Claude and Kognia's Semantic Layer.

Instead of relying on static prompts or manually uploaded datasets, Claude can securely access Kognia's structured tactical database in real time. Every question is translated into the appropriate query against the Semantic Layer, allowing Claude to retrieve accurate information while preserving the consistency of Kognia's data.

The result is an experience that feels less like querying a database and more like having a conversation with an experienced football analyst.

Ask football questions

One of the biggest advantages of the integration is its simplicity. Users no longer need to understand database schemas, metric names or filtering logic. They simply ask the question they want answered.

For example:

“Show me the U23 left full-backs who attack space well, how that translates into a penalty-box threat, and how much space they get from the nearest defender the moment they enter the box.”

Claude understands the football terminology, interprets the user's intent, queries the Semantic Layer and delivers an immediate answer. Detailed analysis using Kognia tactical data become accessible not only to analysts who can bring technical skills, but now also to coaches, scouts and sporting directors.

Explore data beyond traditional dashboards

Dashboards are excellent at answering predefined questions. However, football analysis is rarely predefined. The Claude MCP Integration removes those limitations by allowing users to explore data dynamically. Instead of waiting for a new dashboard or report to be built, users can combine competitions, tactical situations, match contexts, time periods and player profiles simply by asking.

Even better, every answer becomes the starting point for the next question.

A conversation might begin with:

“Show me which winger creates the most off-ball threat, combining how often they occupy dangerous space inside the box with their total volume of off-ball movements and size each point by how marked they are when they get there.”

And naturally continue with:

• Only in LaLiga.

• Under 23 years old.

• During the second half.

Analyse performance in context

Football performance is highly contextual. A player's behaviour changes depending on the opponent, the score, the tactical system or the phase of the match. Looking only at season averages rarely tells the full story.

The Semantic Layer allows Claude to analyse performance across multiple contextual dimensions simultaneously, including match state, competition, tactical phase, time period, player, team and season.

Rather than presenting isolated statistics, the integration helps users understand performance within the context in which it actually happened.

Firstly, “show me which LaLiga strikers press the most and recover the ball most effectively while their team is trailing.”

Secondly, “show me the average ball recovery position for LaLiga strikers on a pitch map, sized by recovery volume, highlighting whoever combines the highest recovery height, pressing volume, and recovery volume; not just wherever the dot happens to land.”

Finally, “for the top 3 strikers by that merit score, show their second-half pressure heat maps side by side, and overlay their actual recovery locations as distinct purple dots. I want to see where they press versus where that pressure actually turns into a recovery.”

Into the detail with Dani Olmo

What makes this example different is scale: instead of a league-wide average, this is one player, one match, event by event. Thanks to Kognia's tracking data exposed through the MCP connector, Claude can reconstruct not just that Dani Olmo received the ball between the lines seven times against Valencia, but where his run started, how fast the pass arrived, how fast he was moving into the space, and how much room he had from the nearest defender when the ball reached him.

A season aggregate like "13.23 receptions per 90" is useful for scouting and benchmarking, but it flattens ninety minutes of very different situations into a single number. The moment you ask to see one match instead, the picture becomes concrete: reception two was a 20.5m first-time ball arriving at 58 km/h with a defender a metre away; reception three, a few minutes later, gave him over five metres of space. Two receptions, same player, same tactic label and completely different situations on the pitch.

Self-service football intelligence

Instead of relying on analysts for every ad-hoc request, football professionals can answer many questions independently.

Scouts can search for players matching specific tactical profiles. Coaches can investigate tactical behaviours before training sessions. Performance analysts can explore contextual metrics in seconds.

The result is a faster, more collaborative workflow where football intelligence is available to everyone.

To illustrate this in practice, we ran a query using the centre-back defensive-line variable and its compactness: "For two centre-backs from the same team playing genuinely opposite sides (one left-footed, one right-footed), for example, Deportivo Alavés's Jon Pacheco and Nahuel Tenaglia show me their pressure heat maps and their average ball recovery position on the pitch. Add a cross at each recovery point sized by how compact their defensive line is in depth and width.”

This is exactly the kind of question that would normally require a bespoke query or a purpose-built dashboard. Here, it took one sentence and returned the underlying data available for immediate follow-up, whether that means splitting by opponent, by game state, or cross-referencing against goals conceded in behind.

To find out more about how Kognia data can be integrated within your club's workflows and AI technology, please email: info@kogniasports.com