I wrote this post as my submission to the August edition of the IndieWeb Carnival, which each month invites folks to write something on their blog based on a topic proposed by a rotating host. August’s carnival topic is “purpose” and the host is Chris Shaw. I am also double dipping and counting this as my first post of Blaugust 2026 - a community celebration of blogging. For the last 15 years or so my job1 has been (more or less) to think about and build the mechanisms to deliver customer support for companies who (more or less) make software on the internet that (more or less) helps other businesses do something more efficiently. In that time I’ve said a lot of things about what the purpose of that labor is. Support as the face of your product. Support as your connection to your customer’s needs. Support as a viable, valuable and under-appreciated career. My understanding of the purpose of the work has shifted and grown alongside me. A photo of myself and the early Zapier support team, taken in August 2015 at our retreat in Traverse City. Zapier was one of my first start up jobs and immensely influential on my philosophy around customer support. The latest evolution of this opinion has less to do with changes in my own view of the work and more to do with changes in the macroeconomic landscape. While the death of support has been confidently decried many times in the past, I believe a time of transformation is really at hand - even if there is life still in the profession. I’m certainly not the first to say that AI - and more precisely LLMs - are dramatically changing both the support customers receive and the daily lives of the people giving it. At no time in the past has it been so easy for a company to give somewhat acceptable answers at such low costs and, debatably, quality. Every recent YC batch is packed to the gills with companies offering the next greatest AI support tool (whose founders largely have never done customer support.) Much ink has been spilled about the death of legacy service leviathans like Zendesk and the ascendant rise of the AI-natives, Decagon or Pylon or whoever. Some of these tools are even good. They leverage sophisticated workflows and data connectors that dutifully and (more or less) accurately answer customer’s endless inquiries. The purpose of a customer support team (or at least the good teams) has never been answering basic customer questions. Those questions are only the most easily quantifiable and measurable expression of customer support. They’re only the shadow on the cave wall of your customer’s experience. The AI tools are built by product managers to automate away the lowly and tedious work of support. They cheerfully deliver answers to these questions, instantly and endlessly, at a quality we’ve never seen before. But designing for this is fundamentally flawed. They mistake the shadow for the underlying truth. They are not replacing the real value of your team. The purpose of your support team, framed in capitalistic purity, is enabling your customers to realize the value of the widget you’re selling. The questions asked are a manifestation of the friction your product adds to this process. Modern software is the monstrous result of groups of people with narrow targets and Quixotic goals competing across the surface of your product. This amalgamation rarely represents what is good or useful for your customers. The purpose of a support team is to know the thing itself and not the shadow. They are tasked to understand customers. Through regular observation and interaction they become the best source on the customer’s actual goals and needs. An understanding not contaminated by a product group’s incentives by way of their myopic metrics. They must understand so that they can communicate and advocate on behalf of the customer’s true needs and experience. My hope is that great teams and products will use the time they gain back from AI handling the rote and rhythmic questions to give their support team’s more time to focus on this higher purpose. Time to write up product findings that reduce customer confusion. Time to sit in on product meetings to offer guidance that steers new developments away from pitfalls. My fear is that most teams will just become smaller teams, still burdened with the reactive work of answering a harder set of questions. If you want to know more about my work history, you can check out the horror show that is LinkedIn ↩︎