Blog

Insights on AI support, multilingual infrastructure, and building on the edge

Product updates, engineering deep dives, and stories from the team building Helpway.

Helpway v1.2 — Smarter RAG, lower latency, new MCP server

Our biggest release this quarter: embedded reasoning traces, 2.4× faster first-token, and an MCP server so Claude Code can read your docs.

AI hallucinations are a product–market mismatch. Here is how Helpway avoids inventing answers entirely.

No translation layer. No per-language agent config. Just one English docset and a visitor who speaks anything.

First-token latency from Tokyo: 620ms. From a single US-East region: closer to 2.1 seconds. Geography is the bottleneck.

71% fewer tickets

A deep look at how the team rolled out a multilingual AI agent across 12 markets in under a week, and what changed.

A short, practical walkthrough: script tag, CSP policy, identity mapping, and how to avoid the 3 most common pitfalls.