AI Agent Engineering Blog
Orloj vs. LangGraph vs. CrewAI: 2026 Update
Six months since our original comparison. All three frameworks shipped major updates. Here's what changed and what didn't.
Jon Mandraki
Why Every Agent System Needs a Governance Layer (Not Just Guardrails)
Guardrails check outputs. A governance layer controls inputs, execution, access, and budget. They solve different problems. Most teams need both.
Jon Mandraki
Orloj vs. Microsoft Semantic Kernel Agent Framework
Microsoft's Agent Framework brings .NET, Python, and Java support with deep Azure integration. Orloj is language-agnostic and cloud-agnostic. Different trade-offs for different teams.
Jon Mandraki
Agent Governance for Financial Services: What Compliance Teams Actually Need
Regulators care about audit trails, model risk management, and data access controls. Most agent frameworks provide none of these. Here's what financial services compliance actually requires from agent infrastructure.
Jon Mandraki
Agent Infrastructure Glossary: 30 Terms You Need to Know
Agent orchestration, tool isolation, fail-closed governance, dead-letter queues. The terminology is piling up. Here's what each term actually means, without the marketing gloss.
Jon Mandraki
Orloj vs. LangGraph vs. CrewAI: When to Use What
Three different tools for agent workflows. LangGraph for stateful conversation. CrewAI for simple teams. Orloj for production operations. Here's an honest breakdown of what each is designed for, and when to use something else.
Jon Mandraki
Agent Cost Attribution: How to Know Which Agent Is Burning Your Budget
Multi-agent systems hide costs. Multiple agents, multiple models, variable token usage, tool calls that cost money. You can't optimize what you can't see. Here's how to set up cost attribution and control agent spending.
Jon Mandraki
Agent Governance for Healthcare: HIPAA, PHI, and the Audit Trail Problem
Healthcare systems running AI agents face a gap most platforms can't close: proving what data was accessed, when, and why. HIPAA requires it. Here's what compliant agent governance actually looks like.
Jon Mandraki
Agent Observability: What to Monitor When Your Software Has Opinions
Traditional APM metrics don't measure what matters for AI agents. Token consumption, decision quality proxies, tool call patterns, governance violations, and cost per task are the metrics that tell you if your agent is actually working.
Jon Mandraki
Top Agent Governance Strategies for Enterprise in 2026
Enterprise teams running multi-agent systems need governance. Here are the six strategies that actually work: policy-as-code, role-based access control, audit trails, rate limiting, human checkpoints, and guardian agents. How each one works, when to use it, and how Orloj implements it.
Jon Mandraki
Orloj vs. Google ADK: Cloud-Native vs. Cloud-Agnostic Agent Orchestration
Google's Agent Development Kit ties into Vertex AI and Gemini. Orloj runs anywhere. Both let you declare agent systems. Here's how to choose.
Jon Mandraki
Fail-Closed vs Fail-Open: Why Your Agent's Default Matters More Than Its Logic
Most agent frameworks default to fail-open: agents can do anything unless explicitly denied. That's a disaster in production. Here's why fail-closed governance matters, and how Orloj enforces it at the runtime layer.
Jon Mandraki
The 2026 Guide to Agent Orchestration: Patterns, Tools, and Trade-offs
A practical guide to orchestration patterns, the tools that implement them, and how to choose the right approach for your agent systems.
Jon Mandraki
Testing Agent Systems: Why Unit Tests Aren't Enough
Unit tests break down when your code doesn't have deterministic outputs. Agent systems need a different testing strategy: property-based testing, contract testing for tool APIs, scenario replay, and chaos testing for resilience.
Jon Mandraki
Top 5 Open-Source Agent Orchestration Platforms in 2026
Orloj, LangGraph, CrewAI, AutoGen, and Temporal all solve parts of the agent orchestration problem differently. Updated for 2026 with Google ADK and OpenAI Agents SDK coverage.
Jon Mandraki
EU AI Act and Agent Systems: What You Need to Do Before August 2026
The EU AI Act's enforcement date for high-risk systems is August 2, 2026. If you operate agents in healthcare, finance, HR, or critical infrastructure, you have four months to implement governance, audit trails, and technical documentation. Here's what needs to happen now.
Jon Mandraki
The Cost of Ungoverned Agents: What Shadow AI Means for Your Engineering Org
When teams build AI agents without governance, they're not moving faster—they're borrowing capacity from your ops team. How shadow AI happens, what it costs, and why good governance actually speeds up adoption.
Jon Mandraki
How to Run Multi-Agent Systems Without Losing Sleep (or Data)
Multi-agent systems fail in ways your existing operational playbooks don't cover. Here's how to handle state management, cascading failures, and task ordering in production.
Jon Mandraki
How to Orchestrate AI Agents in Production
A step-by-step tutorial. You'll build a 3-agent research pipeline with governance, deploy it, and run it. Actual YAML, actual commands, no hand-waving.
Jon Mandraki
Declarative Agent Management: Why YAML Beats Imperative Code for Production
Declarative manifests give you what imperative code doesn't: auditability, drift detection, and GitOps workflows for agents. Why we built Orloj on YAML instead of a Python library.
Jon Mandraki
Orloj vs. Kagent: Declarative Agents Without Kubernetes
Both Orloj and Kagent believe agents should be declared, not coded. But Kagent needs a Kubernetes cluster. Orloj doesn't. Here's when each one makes sense.
Jon Mandraki
The Missing Layer: Why Agent Frameworks Need an Orchestration Plane
Agent frameworks teach you how to build agents. Cloud providers give you compute. But nothing in between handles the operational reality of running agents at scale: scheduling, governance, reliability, observability. That's the missing layer.
Jon Mandraki
What Is Agent Orchestration?
Everyone talks about agent orchestration. Most of them mean different things. Here's what it actually is, why it matters for production, and how to think about choosing an approach.
Jon Mandraki
What Is Orloj? A Technical Overview for Platform Engineers
A deep dive into Orloj's architecture, resource model, and deployment options. For platform engineers evaluating orchestration solutions for AI agent systems.
Jon Mandraki
Guardian Agents Are Coming. Is Your Infrastructure Ready?
Gartner predicts 40% of CIOs will demand guardian agents by 2028. Here's why that matters, what you're actually building, and why bolted-on governance won't cut it.
Jon Mandraki
What Happens When Your Agent Fails at 3am
Agent failures are different from service failures. Your retry strategy needs to be too.
Jon Mandraki
The Governance Gap: Why Most Agent Deployments Fail Before They Start
80% of Fortune 500 companies are deploying AI agents. Only 14.4% have actually gotten security approval to run them. Here's why governance is the real blocker.
Jon Mandraki
Introducing Orloj: Agent Infrastructure as Code
Orloj is an open-source orchestration plane for multi-agent AI systems. Define agents, tools, permissions, and workflows as YAML manifests. Run them with governance built in.
Jon Mandraki