Shopify
Stripe
Slack
Notion
Google Workspace
Confluence
All articles
969 articles
Back to blog home
Build an AI Support Agent with LangChain — and When a Platform Beats a Framework
Build a customer support agent with LangChain and LangGraph — create_agent, tools, and LangSmith for evals — plus an honest look at framework version-churn and the deploy/integrati
No-Code AI Agent: Build a Customer Support Bot Without Writing Code
How to build a customer support AI agent with no code — what no-code can and can't do, the real steps (connect, ground, define tools, test, deploy), and when you'd still need a dev
Build an AI Support Agent in Python (From First Principles)
Build a customer support agent in pure Python — the real agent loop, tool calling, a help-desk API call and retrieval, with code — plus the deploy/monitor/eval work tutorials skip,
How to Build an AI Support Agent with ChatGPT (and Codex)
A developer's guide to building a customer support agent with OpenAI — the Agents SDK, Responses API, function calling and file search in real detail — plus the deploy/monitor/eval
How to Build an AI Customer Support Agent with Claude (and Claude Code)
A developer's guide to building a customer support agent with Claude — the Agent SDK, tool-use loop and RAG in real detail, plus the deploy/monitor/eval work most tutorials skip, a
How to Auto-Resolve "Where Is My Order?" Inquiries With AI
Build an AI agent that auto-resolves 'Where is my order?' tickets — Shopify order lookup, tracking reply, and a clean escalation fallback. Step-by-step.
Process Refund Requests Automatically (Safely) With an AI Agent
Build an AI agent that looks up the payment, checks policy, and processes refunds in Stripe — with a confirmation gate on every money-moving action. Full walkthrough.
How to Measure AI Agent Quality: Evaluation and AI-Judge Scoring
A practical guide to measuring AI agent quality: the metrics that matter, how LLM-as-a-judge scoring works, its blind spots, and how to grade your support agents.
Analysing Thousands of Support Tickets With AI: A Practical Guide
How to analyse thousands of support tickets with AI — manual vs AI methods, what to measure, and a real 57-ticket run inside Macha Studies, start to finish.
The Helpdesk-Agnostic AI Layer, Proven: One Platform, Four Helpdesks
Macha is a helpdesk-agnostic AI layer — the same agents and tools run on Zendesk, Freshdesk, Gorgias, and Front, with real read/write actions, not read-only widgets.
What to Automate First: A Maturity Model for AI Support
A practical maturity model for AI support — six levels from deflect to act, what to automate first, the metrics for each stage, and where teams get stuck.
Knowledge Base for an AI Agent: Designing Sources So It Answers Correctly
How to design the knowledge sources behind an AI support agent so it answers correctly: scoping, single-source-of-truth content, chunking, and freshness.
Page 8 of 81