Progress Report: MCP Server Project
RunSignup continues to build capabilities based on Model Context Protocol (MCP) to support direct authenticated access to RunSignup data from popular AI Chat products like ChatGPT. We currently have a working internal pilot that demonstrates a RunSignup hosted MCP server that accepts natural language requests from ChatGPT for specific data.
Here is what it looks like. After connecting to the RunSignup MCP and authenticating with my RunSignup credentials, I asked it to look up a race called “The Great Donut Race”
I gave it permission to continue and it found the race using the appropriate MCP tool it had available to it. MCP tools are simply capabilities exposed via the MCP Server that have access to RunSignup data and can allow AI models to securely interact with that data.
Then I asked ChatGPT for the results of a particular participant. Again using the MCP Server it located the MCP Tool called search_for_results and returned the data I requested:
It found the results for Joe Runner. Since ChatGPT also has access to public data, I asked it to compare Joe Runner’s race results to national averages:
Wow! Fun! Next I started to ask it questions to visualize Joe Runner’s results – here is a histogram it came up with for the front of the pack:
What Does a RunSignup MCP Mean For Our Customers?
With this example you can start to see how having a natural language conversation with an AI can unlock insights into a data set, such as RunSignup results. AI products will increasingly become a common way participants want to find events, interact with event data (such as results), and even register for an event. This will happen via products such as Google Gemini, ChatGPT, Claude, Meta AI Glasses and a host of other entry points. RunSignup will continue to lead in innovations around providing real-time data to these products for the best end user experience. Our beta MCP server demonstrated above currently has tools for:
get_user_info – gets data about the logged in user
search_for_results – search results within a specific event
search_event_results – search results across events in a race
get_race_id – find races and metadata about races
Future Technical Strategy for the RunSignup MCP Server
Our recent MCP Server prototype is just a tech spike to prove out some concepts. We are working on a complete infrastructure and framework to support AI capabilities that will meet our standards for high security, fault tolerance, and performance. Here is a glimpse of the projects we are working on to support the future of RunSignup (and TicketSignup) in an AI world.
Analytical Focused Data Warehouse
Given the large data sets in RunSignup’s database, it is unlikely that we will want to ship massive amounts of data to AI models for each request. To support analytics rich responses to AI requests RunSignup is investigating alternate large data pools that are optimized for OLAP operations (Online Analytical Processing). These types of data warehouses are much different than traditional relational databases. OLAP data stores are optimized for complex queries and aggregations – and often use columnar storage technologies not relational databases. With this in place RunSignup can do some of the heavy lifting on its current infrastructure prior to shipping response to the LLM for further refining.
Take a request like this:
Look at all my 5k results for the past 10 years on RunSignup and provide a graph of my average times for each year. Provide a reference line on the graph that includes the average 5k times of all men that have run races on RunSignup for that year.
AI models can handle these types of queries, but it would be inefficient to send millions of raw rows of results data to ChatGPT just to calculate this one user’s request. But if we were able to perform analytical calculations against our large data set efficiently on RunSignup infrastructure, we can ship back only the data ChatGPT needs to build a compelling graph.
This type of effort will require understanding the types of queries we want to support, building an efficient data storage system optimized for analytical queries, keeping the data system in-sync with the latest data from RunSignup’s platform, and ensuring we are honoring the access controls put in place in the RunSignup platform.
More MCP Tools
Once we have a better understanding of the data we want to expose to AI models, we will need to build many more specialized MCP Tools that AI models will be able to contact to gain access to the data. With the first four tools completed we are starting to understand the blueprint for this effort.
MCP Server Registries
MCP Registries are how AI tools will find available MCP servers they can use. For our prototype we had to configure ChatGPT manually to know about our RunSignup MCP Server. In the future MCP Servers will be discoverable and AI tools will be able to connect to them seamlessly by looking them up on MCP Registries. Standards are still emerging but the industry organization creating MCP specifications will likely be one of the Public MCP Registries. RunSignup is carefully following MCP Standards so that our products are ready to participate. Among RunSignup’s deep bench of experts are several senior technologists that have navigated technology standards in the past dating back to the emergence of the internet.
Agentic Commerce
Once users are engaged in a discussion with their favorite AI agent to find just the right event for them, they will want to complete the transaction in the same conversational manner that got them to this point. Enter Agentic Commerce, the ability to have AI agents complete an entire registration transaction including the payment. Agentic Commerce is emerging quickly. An Oct 2025 McKinsey study estimates the US B2C retail market could see $1 trillion in transactions occur via agentic commerce. This means an entirely new set of standards, technologies, security guardrails and payment orchestration layers are being invested in to support this shift. RunSignup is a payment facilitator in close partnership with the top-tier payment processors Adyen. Our growth, transaction volume, and eye on innovation has sparked some early discussions with Adyen who is already working with big AI firms and card brands on agentic commerce. RunSignup plans to have working prototypes of race registrations occurring via an AI chat interface in a few weeks.
More Resources
Things are happening fast! You can be sure we are on top of AI industry developments and will report back on what we are learning and building.
For more information on our MCP Server project, register for the upcoming webinar Open API and MCP with founder Bob Bickel and software engineers Duy and Jon on November 19, 2025.
Check the RunSignup AI landing page for all things AI.
Learn by watching – RunSignup AI Video Library