Podium Customer Data Sync & AI Analysis

The problem

The business talked to its customers through Podium: text messages, Google reviews, review requests and appointments. All of that information stayed inside Podium, so they couldn't put it next to their own customer and order data, and nobody could easily see the overall picture.

What I built

A system that copies everything from Podium into the business's own Supabase database automatically:

• New messages, contacts and reviews arrive within seconds of happening.
• Once an hour, it picks up the things Podium doesn't send on its own, like whether a conversation is open or closed, which staff member it's assigned to, appointments, and review replies.
• Every 10 minutes it retries anything that failed. Once a day it emails an alert if something looks wrong, such as no activity coming through or the Podium connection dropping.

AI analysis

With all the conversations and reviews in one place, AI analysis reads through them and shows what is going on: what customers are asking about, what they are unhappy with, and which conversations still need a reply.

Keeping the live database safe

This was a live production database, so I followed strict rules. I didn't change a single existing table. All Podium data goes into its own separate area. The sync uses its own login, which can only write to that area and can't delete anything. Times are stored in the business's local time to match the rest of their data.

The result

Live since October 2026. The team no longer has to dig through Podium. They can ask plain questions like "What are customers complaining about this week?" or "Which conversations are still waiting for a reply?" and the AI analysis answers from real data.

The same setup was later reused for my RingCentral Call Tracking, Transcripts & AI Analysis project.