How customer teams became agent builders
We launched Ghostwriter in March with a simple goal: building and optimizing agents should be as easy as describing what you want. No clicks or code, and no waiting on an engineer. Three months later, people who have never written a line of code (support leads, operations managers, QA teams) are now directly shaping the customer experience.
From handoff to hands-on
The people closest to customers have always had valuable insight into what could be improved. But turning that insight into action often meant navigating a long chain of handoffs and approvals. AI is changing that. The people dealing directly with customers can now make improvements themselves.
A great version of most agents already exists somewhere in a company's conversation logs, buried in call transcripts, support tickets, training documentation, and institutional knowledge accumulated over years. Ghostwriter takes all of that, whether it’s in PDFs, recordings, or zip files, and identifies what's useful, how it should be used, and builds from there.
But building is only part of the story.
The faster an agent goes live, the faster you get feedback. The real challenge is understanding what's important in that feedback, and knowing how to put it into action.
“The biggest thing Ghostwriter has given us is iteration speed. Rather than reviewing conversations, trying to figure out what went wrong, then hoping our fix gets it right, we can just ask Ghostwriter. That alone changes how we work. That speed is what gets customers to a better experience faster. We're not waiting on slow cycles to course correct anymore. We can just move.”
Delan Diaz
Senior Manager, AI-Enabled Operations
Own the agent improvement loop
Enter Explorer, our agent for optimizing agents. It continuously analyzes customer conversations and surfaces the insights that matter. Why are customers dropping off? Which questions isn't the agent handling well? Where is CSAT quietly taking a hit? Instead of requiring teams to hunt for answers, Explorer surfaces what matters and Ghostwriter then helps teams turn those insights into improvements.
The result is a much tighter loop between learning and action.
The best customer experience teams don't let issues sit. They find them, fix them, and move on. Historically, that process involved handoffs between teams. A support lead identified an issue, an analyst investigated it, and an engineer implemented a fix. Everyone waited for the next release.
Today, the people closest to the problem can resolve it directly.
“Ghostwriter has made Agent improvements accessible across our entire team, while allowing us to double-down on our commitment to quality. We care deeply about the customer experience, so we’re constantly reviewing Agent responses and identifying opportunities to improve. With Ghostwriter, team members can now build, test, and launch those improvements themselves in real time. What once required days or weeks of coordination across multiple teams can now happen in real time, dramatically accelerating both our output and impact.”
Mary Orrell
Vice President, Customer Operations
We think this shift is bigger than any individual product feature. The most important change isn't that AI agents are becoming easier to build. It's that the people who often understand customers best are now able to shape the experience themselves. And when more people can build, customers are better off because of it.
