How Atomicwork made help center updates 10x faster with Pageloop

Use case: Help center maintenance Industry: IT service management

Riya Sebastian & Aishwarya Hariharan - Product Marketers at Atomicwork
Pageloop features used: Article creation, Update detection, Proactive suggestions linked to our Slack integration

Atomicwork is an AI-powered IT service management platform for mid-market and enterprise companies. The product covers ticketing, asset management, AI-powered workflows, and enterprise configurations. The company was founded in Chennai in 2022 and ships product changes almost every day.

Riya is on the product marketing team. She and Aishwarya own documentation because the domain is technical enough that writing about it means understanding the product. Both of them used to work at Freshworks, where documentation was central to a self-serve user experience. They brought that standard with them.

"Documentation is very noticeable when it's not there. No one really thinks of it as a time-consuming job. They just notice when it's missing."

- Riya

When every customer had someone to call

Early on, Atomicwork ran a high-touch model. Every customer had someone on the team who was responsible for them. This could be a PM, an engineer, or an implementation specialist. If a customer couldn't find something, they'd ask their person. The help center needed to exist, but nobody depended on it.

As the customer base grew, that changed. Customers started going to the help center first. Internal teams started pointing customers to it too. The help center became the reference everyone relied on. And it wasn't keeping up.

The problem with shipping fast

Atomicwork's engineers post updates across several Slack channels covering product fixes, engineering issues, and new feature launches. A lot of these are small changes that don't get a formal announcement. At one point, a single Slack channel had nearly a hundred updates waiting

Sometimes engineers would work on a batch of small related fixes over days but wait to announce them together. By the time the announcement went out, the changes were already live and customers were already asking. Occasionally, a release would go out, Riya would update the docs, and engineering would roll it back a couple of hours later.

Before Pageloop, keeping up with all of these changes meant Riya reading through articles and relying on memory to find what needed changing. The help center has hundreds of active articles covering ITSM configurations, third-party integrations, and AI workflows. She'd catch the articles she remembered, but naturally there were always some she missed. Customers would find one article saying one thing and another contradicting it.

The team's CTO had built a pipeline to help. Scripts pulled engineer updates from Jira and GitHub into Slack as markdown. A custom GPT converted those into customer-friendly language. Then Riya reviewed each one to strip internal details. Four steps before anyone touched the help center, and articles were still falling behind.

Atomicwork is currently shifting from "AI agents" to "AI co-workers" across the product. This is a single terminology change that will potentially affect dozens of articles which would normally take several hours to fix.

Documentation was taking roughly a third of both Riya and Aishwarya's week. They'd considered building their own system with Claude, but if something breaks, you're debugging an automation pipeline on top of all your other work, and the documentation still isn't done.

How Pageloop changed the workflow

Before Pageloop, most of Riya's time went to two things that had nothing to do with writing. Discovery: figuring out what changed and which articles it affected. And context-gathering: pulling together spec docs, design files, Slack threads, and sometimes sitting with an engineer to understand what was actually built. The writing was the easy part. Getting to the point where she could write was what took hours.

Pageloop handles both. It connects to Atomicwork's Slack channels and watches for changes. When something comes through, it scans the help center and identifies every article that needs updating. It pulls context from the connected signals and produces a suggested rewrite using Atomicwork's existing content, style guide, and terminology. By the time Riya opens it, she's reviewing a near-final draft, not starting from scratch.

A single update run surfaces an average of ten to eleven articles. In early testing, the team rated style accuracy at 95%. About nine out of ten suggestions go through without major changes.

"Someone will say, can you get me this article by tonight? And now I can. Five minutes, done. Before Pageloop, this seemed impossible."

- Riya

What it looks like now

In just their first full month on Pageloop, the team processed nearly a dozen releases and reviewed more articles than they would have gotten to in weeks of manual work. An update that used to take Riya two and a half hours now takes 15 minutes. Today Pageloop has been able to maintain almost the entirety of their help center.

Enterprise customers with complex configurations test changes in sandbox before production. They need documentation before the feature even hits sandbox. That's now possible.

Customer success and implementation teams share articles without checking them first. The complaints about outdated information have dropped. Engineering started announcing small changes more consistently, because they knew docs would pick them up. Better tooling pushed the company toward better habits.

When a new team member joined, they became the highest-volume Pageloop user at Atomicwork. ITSM is a hard domain to onboard into. Writing about ticketing configurations and AI workflows normally takes months of learning. With Pageloop, this new hire was producing articles in the team's voice within their first week. For a two-person team, that turned a training burden into extra output.

With documentation no longer consuming a third of their week, Riya and Aishwarya have time to focus on the creative, more demanding side of product marketing. Launches, enablement, content, website. The work that was always getting squeezed by the help center.

"Creating something from scratch isn't the hard part. It's knowing that what I said in one article needs to match what I've said in ten others. That's where we were losing time, and that's what changed with Pageloop."

- Riya