Testing Tool Execution: Verifying Agent Tool Calls and Side Effects
Learn how to test AI agent tool execution with tool mocking, call verification, parameter assertions, and side effect tracking using pytest in Python.
Agentic AI, LLM engineering, and the models behind modern automation — multi-agent systems, LLM evaluation and comparisons, RAG, fine-tuning, AI infrastructure, security, and production AI engineering.
From the blog
Learn how to test AI agent tool execution with tool mocking, call verification, parameter assertions, and side effect tracking using pytest in Python.
Build robust text classification systems for AI agent routing using intent detection, multi-label classification, and zero-shot approaches with practical Python implementations.
Build robust text preprocessing pipelines for AI agents covering HTML stripping, Unicode normalization, tokenization, length management, and input sanitization with production-ready Python code.
Implement text similarity and semantic matching for AI agents using cosine similarity, sentence-transformers, cross-encoders, and deduplication strategies with production Python examples.
Abstractive text summarization: compare extractive and abstractive summarization methods for AI agents, learn to control summary length and quality, and implement key point extraction pipelines in Python.
A detailed comparison of ElevenLabs, OpenAI TTS, and Play.ht for voice AI agents — covering voice quality, latency, voice cloning, emotion control, and pricing to help you choose the right TTS engine.
Deep dive into building robust text-to-SQL agents that understand database schemas, generate accurate SQL from natural language, validate queries before execution, and gracefully handle errors through iterative refinement.
Implement an AI-powered ticket classification system that automatically assigns priority, department, and SLA to incoming support requests using multi-label classification and intelligent routing rules.
Learn how to store and analyze AI agent time-series data including token costs, latency, and throughput using TimescaleDB, partitioning, retention policies, and aggregation queries.