Part of Bob’s continued ramblings.
This thought has been dawning on me the past several months. AI is introducing a new user experience to traditional transactional and database driven systems with Chat interfaces, Vibe Coding, and Agents. This is part of a continuing evolution from dumb terminals -> PC applications -> GUI point and click applications -> Web applications -> Mobile applications over the decades and is the next major move forward for applications like RunSignup.
AI’s Next Step – Making Databases Easy (and Secure) to Access with Natural Language
So far most of the talk around AI is people prompting Large Language Models like ChatGPT and Anthropic’s Claude. These AI LLM’s are based mostly on static content they have ingested from the web or text.
The next large set of content to ingest is held in databases. The AI startup companies know that database knowledge is very valuable and represents a huge market either selling to enterprises, or powering a new generation of SaaS business applications. This is why there is a huge rise in Vibe Coding tools like V0 and Replit as well as an increasing number of agent based vertical applications.
The big LLM companies like OpenAI and Anthropic also recognize this opportunity. Anthropic nailed an approach in November, 2024 called Model Context Protocol (MCP). This provides a standard mechanism for database and transaction based systems like RunSignup or Salesforce to provide secure access to the functionality and data held in these systems. MCP is being rapidly adopted by all the LLM’s and many, many software vendors.
The transactional side remains dominant and important because of the need to know exactly who signed up for a race, exactly how many people are coming to each ticket time slot, and exactly how much money has been collected. No room for hallucinations. However, the AI side opens up a new interactive mode for these systems that make them more approachable by a much wider audience. And with the help of AI, many tasks will become easier and easier we humans gain a tutor and data sifter and analysis co-worker who will be increasingly capable.
What Can Be Done with MCP-Enabled Database and Transaction Systems
MCP is being adopted rapidly by all the LLM’s and many third party application developers like RunSignup. We have beginning MCP servers running in development now that allow natural language interface to our system and data. For example, we run Claude in our development environment and it is configured to talk with the as yet not released RunSignup AI MCP to ask questions like:
“I want to run a race this weekend, but I don’t want to spend over $25.”
“Are there any races next weekend where I can bring my baby stroller and walk the course.”
Under the covers, the LLM is understanding the meaning of the request, passing it to the MCP server, which then gets translated into API calls that return all the content the LLM can process and understand. Our MCP is providing access to all the public race information on all of the local race websites, including the data like price and distance and location coming from our database. The LLM then answers with a list of options that meet the criteria for the user. In some ways, this is the same as going to the Find a Race page on RunSignup. But it is better because it is a natural language interaction and is based on much more data like FAQ’s containing information like if baby strollers are allowed.
We also have a little experiment to learn more about MCP’s and have hooked up a Salesforce MCP server to Salesforce using Claude Desktop. After providing secure login via OAuth2 to Salesforce, we are able to query our own Salesforce system asking questions like:
“Give me all the customers in New Jersey who processed over $10,000 of transaction volume in the past 3 months and have their event happening in the next 30 days.”
For anyone who has tried to navigate Salesforce reporting, you will know that many users will welcome a natural language user interface.
RunSignup AI
RunSignup AI is made up of a number of projects and products we will be rolling out incrementally. The goal of RunSignup AI is to enable that diagram above. We will provide:
- Core services like Open API, Oauth2, MCP as well as roadmaps for customers to use AI alongside RunSignup’s Platform.
- We will provide practical AI tools, chatbots and agents that event directors can use to enhance the customer experience for their customers.
Our first big release is this coming week (June 24, 2025) at RunSignup University.
The Future
We envision a future where event directors will be able to chat with RunSignup to understand their event better:
- “How are we doing compared with last year?”
- “How many shirts and what sizes should I order this year?”
- “How many people are coming between noon and 3 on Saturday?”
Event participants and attendees will be able to:
- Use tools like ChatGPT to find events.
- Sign up for events (imagine just like Apple has ApplePay, ChatGPT has ChatGPT Pay).
- Manage registrations and ticket purchases with natural language to do things like transfer events, change times and transfer tickets to friends.
We look forward to working with our customers to realize the full potential of AI for events.