Everyone wants to build an AI agent right now.

Not just a chatbot. Not just a prompt wrapper.

A real AI agent — something that can understand goals, use tools, remember context, interact with users, and improve over time.

…[B]uilding an agent is not one decision. It’s a system design problem. An AI agent only becomes useful when several layers work together — purpose, prompts, models, memory, orchestration, interfaces, and evaluation. — Read More

#devops

In practice, when building agentic systems, AI models are rarely the bottleneck anymore. The harness around them is. Anthropic spent two years building that harness for Claude, the orchestration code that picks the right tools and grades its work before declaring success. Claude itself is built around 62 carefully composed components spanning machine learning patterns like compute optimal allocation, deliberative alignment, bi temporal memory, alongside agentic patterns like the OODA loop, plan and execute, architect editor splits, and many others.

Those 62 components that define Claude’s thinking approach are distributed across 4 main principles: Cognition, Orchestration, Reliability, and Grounding and Trust. — Read More

#devops

What did I build this week?

An email app…

I use Gmail. I’ve used Superhuman for years. I like it a lot. It is fast, keyboard-first, clean, and is good software. But like many saas products, it keeps adding features that I don’t need and more importantly, I don’t need to be paying for email.

I wanted a split inbox and rules to organize my emails.

Kicking off with Codex: — Read More

#devops

Most AI agent failures do not trace back to bad reasoning. The model understands the task, then calls the wrong tool, passes malformed arguments, gets back an unhandled error, and produces a wrong answer anyway. The reasoning layer gets the attention; the tool layer is where production incidents actually happen.

Tool calling — also called function calling — is what bridges a language model’s reasoning to real-world action. Without it, agents are capped by training data: no live queries, no external systems, no side effects. With it, an agent can search the web, call APIs, run code, retrieve documents, and trigger transactions in any system that exposes an interface.

Getting this right means understanding the full stack, not just the happy path. — Read More

#devops

I have been building with AI agents and MCP tools for a while now, and one question kept coming up: how do you give an agent real, authenticated access to AWS without handing it the keys to the kingdom? Today, there is an answer.

I’m happy to announce the general availability of the AWS MCP Server, a managed remote Model Context Protocol (MCP) server that gives AI agents and coding assistants secure, authenticated access to all AWS services through a small, fixed set of tools. — Read More

The AI era demands a fundamental shift in security, and that includes identity and access management (IAM). Traditional controls simply aren’t built for autonomous AI agents that interact with sensitive data at machine speed, a reality we address with our new IAM advancements for the agentic enterprise era.

Engineered as built-in Google Cloud capabilities to secure the rapidly-expanding world of AI agents, at Google Cloud Next we introduced a new security and governance paradigm for managing agent identity and access. This comprehensive framework focuses on foundational Agent Identity and an Agent Gateway with Identity-Aware Proxy, while integrating robust agent access management, agent guardrails, and runtime defense to enable a secure cloud environment for your organization. — Read More

A  language model on its own is stateless. You feed it a prompt, you get back a continuation, and the moment the response is finished the model forgets you ever existed. There is no “previous conversation” living inside the weights.

An agent, by contrast, is the orchestration around the model: a loop that decides what context to pass in next. Memory is the part of that loop that carries information forward. Everything in this essay is a different answer to the same question, what should we put in the prompt this time? — Read More

Is it best to use an LLM with its native harness (like Claude Code or Codex), or a generic harness that swaps models on demand?

… [I] decided to dig deeper by looking at the harness implementations of Codex, Claude Code, and Github sdk. Does the harness really matter that much?

… The hand wave answer is that “models behave differently because they are different models.” but here I tested the same models and different harness. — Read More

#devops

Something shifted in the first quarter of 2026. Not a feature launch, not a new product – a structural change in how work happens.

For the first time, I found myself genuinely operating with agents across every dimension of my work: personal tasks, software engineering, company operations. Not as a novelty. As the default mode.

This post is the abstraction I arrived at after weeks of doing this. A mental model that applies everywhere – because the architecture underneath is always the same. — Read More

#devops

Perplexity’s frontier agent products rest on a foundation of know-how and domain expertise packaged in modular Agent Skills. We maintain a carefully curated library of Skills across our technical environments. These Skills include many of the general-purpose utilities powering Perplexity Computer; vertical-specific capabilities in areas such as finance, law, and health; and a very long tail of modules for addressing user needs. Some Skills are infrequently invoked but critical when invoked. To ensure a consistently excellent user experience, Perplexity’s Agents team prioritizes Skill quality just as much as code quality.

The intuitions and best practices required to develop a high-quality Skill differ significantly from those required to build traditional software. The Agents team reviews many pull requests from excellent engineers who develop Skills in the course of their work. The result is almost always numerous comments and suggestions for revision. This is because many useful patterns for writing code become antipatterns in Skill creation. — Read More

#devops