Build AI Agent for Architectural Code Compliance with HATZ AI

Subscriber's AI agent idea

The host introduces the video's focus, which is a subscriber's idea for an AI chatbot to check architectural site codes before design submission.

Building agent with HATZ Workshop Assistant

The host explains how he used GPT-5 to generate a system prompt for the architectural agent and then utilized the HATZ Workshop Assistant to consolidate recommendations and create the agent.

Configuring LLM and web tools

The host selects GPT-5 as the LLM and adds web search tools like Firecrawl and Tavily to enable the agent to access real-time information and regulations.

Testing the architectural code agent

The host demonstrates the agent's functionality by asking it to find building and energy code additions for a specific location and review a brief for a local cafe.

Agent's continuous conversation capability

The host highlights that the agent retains its persona and instructions, allowing for ongoing conversations and complex tasks without needing repeated prompts.

Human in the loop importance

The host stresses the necessity of human validation for AI-generated results, even if the AI is highly accurate, to double-check and verify information.

Blueprint analysis and future capabilities

The host shows how the agent can extract parameters from a blueprint and hints at future capabilities like agents running workflows and apps within HATZ.