Grade the Answer, Not the Query: How We Built MindzEval to Evaluate Our Text-to-SQL Agent
We built an agent that turns plain-English questions into SQL, and it worked. Then someone asked the one question it…
Read More
From ReAct to Deep Agent: When One Routine Is Not Enough
Our chatbot had one way of thinking, and it used that one way for everything. This post is about the…
Read More
Returning Only What Matters: Smarter Web Crawling with Semantic Search
When our agent needed to answer a complex question, it could not rely on search results alone. Sometimes the answer…
Read More
Crawl4AI vs SearXNG: Choosing the Right Web Search Tool for AI Agents
At a glance: Our AI agent's first web search tool used Crawl4AI, which opened a new headless Chromium browser for…
Read More