Blog
Thinking behind the product
Engineering depth from the team building BuildWright.


Build the Agent Around the Domain
We set out to build an AI engineer for support. The hard part was not the AI. It was the domain: which work must happen the same way every time, and which needs judgment.
Read the article
Zendesk and ServiceNow Escalations to Engineering: Where the Time Goes
A stage-by-stage look at what happens when a Zendesk or ServiceNow ticket escalates to engineering, where the hours actually go, and which fixes shorten each stage.

You Can't Hire Your Way Out of a Knowledge Problem
Every conversation with a support engineering leader eventually reaches the same sentence. "We need two more engineers. Otherwise we won't hit SLA."

AI Solved Coding. It Hasn't Touched Investigation.
The #2 enterprise AI use case has no dominant player. Here's why.

Why Your AI Support Bot Is a Chatbot in Disguise
Every enterprise vendor slapped "AI-powered" on their help desk in 2025. But asking an LLM to summarize a ticket is not agentic support. It's autocomplete with extra steps.

The Largest Invisible Line Item in Your Engineering Budget
Why 70% of every support ticket's resolution time is spent searching for the problem, not solving it

From Vector Memory to Context Graphs: The Evolution of Agent Memory
Most agent discussions focus on models and prompts. The real constraint is how the system remembers. A look at how memory architectures are evolving from vector retrieval to context graphs.