Keeping AI Support Human at Paperlike

Building AI agent guidance, support macros, and workflow rules that stay accurate and brand-safe

Keeping AI Support Human at Paperlike

Keeping AI Support Human: AI Support Operations at Paperlike

AI can draft a customer support reply in seconds. Making sure that reply is accurate, compliant, brand-safe, and genuinely helpful — that’s where the real work happens. At Paperlike, I work as a freelance Customer Support & AI Support Operations Specialist, sitting exactly at that intersection of AI tooling and human judgment.

Teaching the AI How to Help

Paperlike uses Gorgias with an AI layer for customer support, and my job is to make the AI a better teammate:

  • AI agent guidance: Writing the instructions that tell the AI how to handle each scenario — when to answer, what tone to use, and when to hand off to a human.
  • Support macros: Creating and refining reusable responses so answers stay consistent, accurate, and easy for agents to personalize.
  • Workflow rules: Building the routing logic that gets each ticket to the right resolution path.

Documenting the Policies Behind the Answers

An AI is only as reliable as the policies it’s trained to follow. I documented the support knowledge that powers both human and AI responses:

  • The 100-day Satisfaction Guarantee and how it applies in edge cases.
  • Opened-package returns, replacement eligibility, and refund workflows.
  • Amazon and retailer purchase flows, tax-exempt customers, and shipping delays.

The Unglamorous (But Critical) Side: Logistics

Great support also means knowing what happens after someone clicks “buy.” I built operational knowledge around DHL, MailAmericas, PostNord, DPEX, and Floship, customs processes, tracking stagnation, and country-specific delivery expectations — so when a package gets stuck at customs, the response is specific and useful, not generic.

Quality Control for AI Replies

Before AI-generated replies reach customers, I review them for tone, compliance, accuracy, escalation needs, and policy alignment. The goal isn’t to replace human judgment with AI — it’s to use AI to deliver faster answers while keeping a human standard of care.

Lessons Learned & Takeaways

AI support operations is technical writing’s newest frontier: the “reader” is sometimes an AI agent, and the documentation becomes the AI’s behavior. Writing clear policies, precise workflows, and unambiguous guidance has never mattered more.