I'm a co-organiser of the "Data & AI Community Day Durban" events. Juggling the practical challenges of organising events can be rewarding. Still, I'll admit that managing contacts and logistics comes with its share of headaches.
I want to develop an AI-driven contact and event management system, helping managing contacts and events.
The goal is to efficiently manage contacts, streamline communications, and generate actionable reports for our events.
Brevo (formerly Sendinblue) has been our event and contact management system for our "Data & AI Community Day Durban" events for a couple of years now, and it has been satisfactory. However, how we want to do things may not match fully how Brevo does it, so limitations have become apparent:
- Workflows are too generic for our events.
- Limited flexibility for managing contacts and participants.
- Reporting lacks the necessary insights for our needs.
The limitations above are some of the reasons I'm developing a new system.
It's not a full-featured event management platform—we're not replacing Eventbrite or building registration systems. What we need is something more focused: a "Contact Relationship Management (CRM) and Communications Hub" specifically for the Data & AI Community Day Durban events.
Here's the reality of my workflow:
- Attendee registration happens through Quicket (a third-party ticketing platform)
- Speaker submissions happen through Sessionize (a speaker management platform)
- Check-ins happen through Quicket on the day of the event
- Everything else currently happens through Brevo or manual spreadsheet juggling.
Attendee registration and check-in on thday, as well as registration of speakers will still be done by Quicket and Sessionize respectively. The system will replace the Brevo/manual spreadsheet juggling.
What I need is a system that:
- Stores contacts: A unified database of all attendees and speakers with their participation history
- Tracks participation: Who registered for which event, who actually attended, who spoke at which event.
- Speaker grading: What speaker spoke abvout what topic(s) at an event, and how was he/she rated, in terms of topic as well as presenting
- Handles communications: Email workflows for speakers and attendees, pulling contacts from our database
- Provides insights: "Show me speaker retention", "How many first-time attendees did we have?", "Which topics had the highest attendance?"
- Imports data: Batch upload registration data from Quicket, speaker data from Sessionize, speaker ratings from an internal system.
Initially I thought about building a traditional API driven system, with a web front-end, etc. and Claude Code would help me to build it. Since then I have changed tack, and my vision is now to build a system that is completely AI driven with LLM's as well as MCP Servers, etc.
What I want is to be able to ask questions like:
Me: "Is Jane Smith registered for the upcoming conference?" System: "Yes, Jane Smith (jane.smith@example.com) registered on December 15, 2024. She also attended the 2023 and 2024 conferences as an attendee."
Me: "How many people attended last year's event compared to 2023?" System: "2024 had 287 attendees (92% of registrations). 2023 had 245 attendees (88% of registrations). That's a 17% increase in attendance."
Me: "Who were the 5 top performing speakers at the latest event" System: "The three tops speakers were: John Doe, Allan Pead, and Kalreen Govender.
Me: "Send an email to all 2024 speakers announcing the call for papers for 2025." System: "I've queued an email to 23 speakers from 2024. Would you like to preview the email list before sending?"
Can you think hard about this and determine whether my idea around LLMs and MCPs, etc., is valid. Just look at it from a high level, do NOT decide on a tech stack yet, that will come later. Just propose a very high level architecture. Oh, the one tech choice is that, if a backend database is needed, it would be PostgreSQL.
"How many people attended last year?" and have the system answer. I also want to send emails to specific groups like "all speakers from 2024". What's the best architecture for this?