The Problem

Each system had its own view of partners. Salesforce tracked sales activity. PlanHat tracked customer success. Customer.io handled marketing automation. Support lived in Intercom, then Dixa, then back to Intercom. None of them talked to our in-app data.

People were carrying information mentally from system to system, or not at all.

The Structural Challenge

In TalkingPoints, a school can exist as multiple instances—different years, different programs. In Salesforce, there's one record per real-life school. These structures don't map cleanly onto each other. Connecting them without major manual work required getting creative.

Fuzzy Matching with Human-in-the-Loop

I built a fuzzy matching system using Jaro-Winkler similarity plus partner metadata. The goal wasn't full automation—it was minimizing friction for humans in the loop. Perfect matches? The human just approves. Close matches? A ranked short list instead of scrolling through every school in the system.

The cognitive load drops significantly.

The Philosophy

This reflects how I think about tech-enhanced workflows (and now AI-enhanced workflows): not "replace the human" but "reduce the friction."

Leaders shouldn't have to carry information mentally from system to system. Connect the silos so they can see what's happening across all of them.

The same principle applies to AI. I'm not trying to automate humans out of the loop. I'm trying to make the loop less exhausting.

The Integrations

Salesforce

Fuzzy matching connected TalkingPoints usage data to Salesforce accounts. The sales team could see which prospects were already using the product organically—context they didn't have before.

PlanHat

Full Snowflake-to-PlanHat integration. Customer Success could see product usage alongside their relationship data. Health scores based on actual engagement, not guesses.

Customer.io

User verification logic to prevent non-verified users from entering marketing automation. Engagement metrics flowing into campaign targeting.

Intercom

Support lived in Intercom, then we migrated to Dixa, then back to Intercom after they announced Fin AI. Integrating Fin into our support workflow was a huge time saver—the same philosophy as everything else here. Let the AI handle the repetitive questions so humans can focus on the work that actually needs them. Enhanced support card UI so agents could see context without switching systems.

Impact

• 4 major system integrations connected to TalkingPoints data
• 360-degree customer views across Sales, CS, Marketing, Support
• Reduced cognitive load for cross-functional teams
• Fuzzy matching framework reusable for future integrations

Technologies