文章介绍如何将“改善”理念应用于 AI 时代的客户支持，通过持续进行小幅优化，逐步提升 AI 驱动支持的效果、流程与服务质量。
Imagine improving something by just a tiny fraction every day.
On its own, that might not sound very impressive, but over time, those small improvements can result in meaningful change.
That’s the philosophy behind Kaizen, the Japanese concept of continuous improvement. Developed in post-war Japan and adopted by companies like Toyota, Kaizen is built on a simple idea that small, steady changes lead to significant long-term results.
However, Kaizen is not simply hoping for gradual improvements over time; it is a structured way of thinking and working, grounded in three key principles:
- Small changes reduce resistance: Big goals can feel overwhelming, but focusing on one small improvement at a time can make it easier to take action.
- Improvement is continuous, not occasional: Instead of waiting for quarterly reviews or major releases, you ask: “What can we improve right now?”
- Everyone participates: The people closest to the work are best positioned to improve it.
In practice, Kaizen follows a simple cycle: identify a small problem, test the change, measure the result, refine, and repeat. The focus isn’t on making radical changes, just steady progress.
At Intercom, we apply this same philosophy to how we manage our Agent Fin through a process we call the “Fin Flywheel”. Here’s how this works.
Building improvement into our everyday use of Fin
Like Kaizen, the Fin Flywheel builds improvement into everyday work through four repeatable steps:
- Train: Teach Fin how to handle and resolve the most complex customer queries.
- Test: Run fully simulated customer conversations from start to finish to see exactly how Fin will behave before going live.
- Deploy: Launch Fin across all channels so customers get consistent support wherever they reach out.
- Analyze: Use AI-powered insights to review and improve Fin’s performance so it can deliver better customer experiences.
This isn’t a one-time setup, it’s a continuous loop where every interaction feeds ongoing improvement. Rather than deploying AI and assuming it will perform as expected, improvement is built into the system itself. The more Fin is used, the better it gets.
Using human insight to continuously improve our AI-powered service
But continuous improvement doesn’t stop with AI.
Within our Human Support team at Intercom, continuous improvement is part of every frontline rep’s role. We operate with a simple mindset: the first time that you solve a customer issue should be the last time it happens.
When a conversation reaches a human, we ask ourselves questions such as:
- Why did this reach me?
- Why couldn’t Fin resolve it?
- How can we prevent this from happening again?
As teammates handle customer conversations, they’re encouraged to identify ways to improve the experience – whether that’s a content update, a process fix, or a tooling gap.
We’ve built a lightweight, AI-powered way to log suggestions in the moment. No long explanations or heavy admin is required. Ideas are reviewed quickly and implemented by subject matter experts or by the team themselves.
This turns our frontline support team from reactive problem-solvers into a proactive improvement engine. The people closest to customers can spot friction, suggest fixes, and see their insights shaped into meaningful change.
Kaizen demonstrates that lasting progress doesn’t come from occasional transformation; it comes from intentional, everyday refinement.
The Fin Flywheel applies that philosophy to AI. Our Human Support continuous improvement process applies it to human insights. Together, they create a shared system where both people and AI learn continuously from customer interactions.
When improvement is built into the mechanics of how you work, it stops being a one-off project and becomes an ingrained capability. Over time, those small daily improvements don’t just add up – they result in competitive advantage.