Why our founders left their jobs to build Pageloop

Every few months, a new AI support tool launches with the same pitch: plug it in, watch your ticket volume drop. And every few months, the same teams try it and get the same result. The chatbot hallucinates. Resolution rates look inflated. Customers get angrier than before.

The usual conclusion is that the AI does not work yet. But that is not what is actually happening. The AI works fine. The problem is that the knowledge base it is pulling from was last updated before your product's most recent redesign. Maybe before the one before that. We have talked to all kinds of knowledge teams - some with regularly managed help centers and some without. The chatbot was doing exactly what it was told. It was just told the wrong things.

This is the problem Nivedha kept running into over nearly a decade of working in customer support at Zapier and Postman. The tools were fine. The teams were trying. But nobody had built infrastructure for what happens when the product moves and the docs do not follow. And now that AI chatbots depend on those docs, the gap has consequences that customers actually feel.

Nishanth came at the same problem from the other side. As an engineer building with AI, he kept watching models get blamed for failures that had nothing to do with the model. The context was wrong. The source material was stale. The AI was a scapegoat for a maintenance problem that predated it by years.

They sat down to talk about why they left what they were doing to build Pageloop, what they have learned from the teams they work with, and a chicken-and-egg problem that keeps repeating across the industry: leaders want to see AI work before they fund it, but AI cannot work without that investment going into the knowledge layer first.

Learn more about Pageloop.