The LayBuild blog
How AI support agents work, and where they break
We build an AI agent that answers support questions from your own documents, says so when they don't cover the question, and lets your team take over the chat. These posts explain how that works, where it falls short, and how to measure it honestly.
1 to 9 of 9 articles
How LayBuild works
Regression-testing a support agent with bun test
Two test layers for an AI support agent: fast contract tests with a stubbed model on every commit, and a golden question set against a real model.
10 min read
How LayBuild works
Adding the LayBuild widget to a Next.js App Router site
How the LayBuild chat widget loads (script, iframe, domain check) and how to add it to a Next.js App Router app with next/script and verified identity.
8 min read
How LayBuild works
One conversation model for web chat, WhatsApp and the API
How LayBuild stores web widget, WhatsApp and REST API conversations in one model, where the channels behave differently, and what it does not support.
7 min read
How LayBuild works
How LayBuild remembers a conversation, and what it forgets
LayBuild's real memory design: 12 messages loaded and trimmed to 6, long-term customer facts, past-conversation excerpts, and the trade-offs of each.
6 min read
How LayBuild works
Event-Driven Customer Support Architecture: Webhooks & Queues
How to drive customer support workflows from webhook events: what the sender guarantees, deduplication, queues, and a receiver for routing and alerts.
8 min read
How LayBuild works
Several focused agents or one: routing in LayBuild
When splitting support into several AI agents helps, how LayBuild assigns a conversation to one agent, and why we have not built an LLM supervisor router.
7 min read
How LayBuild works
API tools without function calling: how LayBuild calls APIs
LayBuild's API tools run fixed HTTP calls instead of letting the model choose. Why, where that falls short, and what real function calling needs.
8 min read
How LayBuild works
Autonomous AI Ticket Resolution: Real Rates, Deflection vs Resolution & Benchmarks
What autonomous AI ticket resolution means, how to measure it honestly, and why many published rates count deflection as resolution.
10 min read
How LayBuild works
AI Chatbot vs Agentic AI: Architectural Differences & ROI (2026)
Compare scripted chatbots, retrieval-grounded LLMs, and agentic AI. Architecture trade-offs, pricing models, failure modes, and ROI benchmarks.
10 min read