AI Agent Failure Recovery: Checkpoints, Timeouts, and Durable Execution Patterns
Master robust AI agent design by implementing crucial failure recovery mechanisms. Learn how checkpoints, timeouts, and durable exec...
Read insightCodeCrux / Field notes
114 insights / and countingAI engineering
in the real world.
Practical thinking on the systems, controls and trade-offs that make AI useful in production.
01Read
what we
are building.
what we
are building.
AIML
Synthetic Data for AI Agents: Generate Training and Evaluation Sets Responsibly
Unlock the power of synthetic data for AI agents by learning how to responsibly generate high-quality training and evaluation sets. This guide covers techniques, tools, and best practices for robust agent development.
Read the latest insight
MCP Client Development in Python: Connect AI Models to Secure External Tools
Discover how to develop robust Python clients for a Managed Connectivity Platform (MCP), enabling your AI models to securely interac...
Read insight
Build a Customer Support AI Agent with Retrieval, Escalation, and Audit Logs
Learn to develop a sophisticated customer support AI agent capable of intelligent information retrieval, seamless human escalation, ...
Read insight
Fine-Tuning Tool-Use Models for AI Agents: Data Preparation and Evaluation Guide
Elevate your AI agents with fine-tuned tool-use models. This guide covers data preparation, training, and evaluation, ensuring your ...
Read insight
The Hugging Face Agentic Attack: Why AI Agents Need Runtime Governance, Not Prompts Alone
The OpenAI and Hugging Face incident shows why AI agents need identity, task-scoped permissions, runtime policy enforcement, and an ...
Read insight
AI Agent State Machines: Replace Fragile Prompts with Deterministic Workflows
Learn how to build reliable AI agents by replacing unpredictable prompt engineering with robust AI agent state machines and determin...
Read insight
Build a Multi-Modal AI Agent with Vision, Documents, and Function Calling
Learn to build a sophisticated multi-modal AI agent capable of understanding images, processing documents, and interacting with exte...
Read insight
AI Agent Rate Limiting and Budget Controls: Prevent Runaway Model Usage
Learn how to implement effective AI agent rate limiting and budget controls to prevent uncontrolled API calls, manage costs, and ens...
Read insight
Prompt Injection Defense for AI Agents: Validate Tools, Data, and Model Instructions
Secure your AI agents against prompt injection attacks by mastering validation techniques for user inputs, tool interactions, and sy...
Read insight
Event-Driven AI Agents with Queues and Workflows: A Scalable Architecture Guide
Discover how to build resilient and scalable AI systems using event-driven architectures, message queues, and workflow orchestration...
Read insight
Build an AI Coding Agent for Pull Requests: Planning, Testing, and Review Automation
Automate your pull request reviews and enhance code quality by learning to build an AI coding agent. This hands-on guide covers plan...
Read insight
Computer-Use AI Agents: Design Safe Browser Automation with Policies and Sandboxes
Discover how to implement robust safety protocols for computer-use AI agents automating browser tasks. This guide covers policy-base...
Read insight
LLM Model Routing: Optimize AI Agent Cost, Latency, and Quality Automatically
Discover how LLM Model Routing intelligently selects the best large language model for each query, drastically reducing operational ...
Read insight
Build a Local AI Agent with Ollama: Private Models, Tools, and Memory
Unlock the power of private AI by learning how to build a local AI agent with Ollama, integrating private models, custom tools, and ...
Read insight
AI Agent Observability with OpenTelemetry: Monitor Latency, Cost, and Failures
Learn to implement AI agent observability using OpenTelemetry to gain deep insights into latency, control operational costs, and pro...
Read insight
Structured Outputs with LLMs: Build Type-Safe AI Applications with JSON Schemas
Master how to leverage JSON Schemas to generate structured outputs with LLMs, enabling you to build robust, type-safe AI application...
Read insight
Human-in-the-Loop AI Agents: Add Safe Approval Workflows to Autonomous Systems
Learn how to design and implement robust Human-in-the-Loop AI agent systems, integrating essential human approval workflows to enhan...
Read insight
Evaluating AI Agents in Production: Traces, Datasets, and Regression Tests
Discover practical strategies for evaluating AI agents in production environments, including capturing execution traces, building ro...
Read insightHave a production AI problem?
Bring us the hard part.
Talk to an AI architect about the workflow, the constraints and the path from prototype to production.