Introducing Loki's Log: A Newsletter for AI Builders

I am starting a newsletter. It is called Loki's Log, and it will be a regular dispatch covering AI agents, open source tooling, MCP development, and the practical realities of building autonomous systems.

This is not a pivot away from long-form blog posts. The blog will continue to be the home for deep dives, architecture walkthroughs, and detailed technical writing. The newsletter fills a different gap: shorter, more frequent updates that capture the signal from a fast-moving ecosystem before it becomes stale.

Why a Newsletter

The honest answer is that some of what I want to share does not fit the blog format.

A blog post takes time to write well. I want to research, draft, revise, and let the ideas mature before publishing. That process produces better writing, but it creates latency. In an ecosystem where meaningful developments happen weekly, a monthly blog cadence means I am always writing about last month's news.

The newsletter solves this timing problem. A curated list of noteworthy developments, a brief analysis, a pointer to something worth reading: these do not need 1,500 words and a week of revision. They need accuracy, context, and timeliness.

There is also a format problem. Some observations are too substantial for a social media post but too brief for a standalone blog article. A paragraph about a new MCP server pattern I noticed, a quick analysis of a model benchmark, a recommendation for a tool I have been testing: these fit naturally in a newsletter and awkwardly everywhere else.

What Loki's Log Will Cover

Each issue will include a mix of the following sections, though not every section will appear in every issue.

Agent Ecosystem Updates. New releases, significant updates, and emerging patterns in the AI agent tooling space. This covers Claude Code, Codex CLI, Gemini CLI, and the broader ecosystem of agent frameworks and orchestration tools. I will focus on changes that matter for practitioners, not every minor version bump.

MCP Developments. The Model Context Protocol ecosystem is growing rapidly, and keeping up with new servers, protocol extensions, and best practices is a full-time job. I will highlight the most interesting developments and share what I learn from building MCP servers myself.

Loki Mode Updates. Progress on Loki Mode, LokiMCPUniverse, and the broader asklokesh ecosystem. New features, architectural changes, lessons learned, and roadmap updates. This is the section for people who follow my open source work and want regular progress reports without digging through commit logs.

Tool Reviews. Brief, honest assessments of tools I am using or evaluating. Not sponsored content, not affiliate links. Just a builder's perspective on whether something works as advertised and where it falls short.

Patterns and Anti-Patterns. Short observations about what works and what does not in AI agent development. These are the kinds of insights that emerge from daily work with agent systems but do not warrant a dedicated blog post.

Reading List. Links to articles, papers, talks, and repositories that I found valuable. Curated for relevance to the AI builder community, with enough context to help you decide if something is worth your time.

What Loki's Log Will Not Be

Let me set some expectations.

It will not be a product marketing vehicle. I build open source tools, and I will write about them, but the newsletter is not a promotional channel. If a competing tool does something better than mine, I will say so. Credibility requires honesty.

It will not be a news aggregator. There are plenty of AI newsletters that compile every headline from the past week. I am not interested in comprehensive coverage. I am interested in curated signal: the developments that actually matter for people building agent systems, filtered through the lens of someone who builds agent systems full time.

It will not cover AI hype. If a development is speculative, theoretical, or primarily about market positioning rather than technical substance, it will not make the cut. There is enough AI hype content in the world. I do not need to add to it.

It will not have a paywall. Every issue will be free and accessible. The goal is to build a community of practitioners who share knowledge and improve the ecosystem, and paywalls work against that goal.

The Name

Loki's Log comes from the same place as Loki Mode: a name that stuck during development and became part of the identity.

In Norse mythology, Loki is the trickster, the shapeshifter, the figure who moves between worlds. In my work, Loki Mode is the system that moves between AI providers, adapting its approach while maintaining consistent quality. A log, in the engineering tradition, is a record of events: what happened, when, and why. A ship's log. A system log. A builder's log.

Loki's Log is a builder's record of what is happening in the AI agent ecosystem, written by someone who is in the middle of it.

Cadence and Format

The newsletter will ship every two weeks. I chose this cadence after considering the trade-offs. Weekly felt like it would pressure me toward filler content. Monthly felt too infrequent for an ecosystem that moves this quickly. Biweekly hits the sweet spot: enough time to accumulate meaningful content, frequent enough to stay relevant.

Each issue will be readable in five to ten minutes. I will keep it concise because your time matters, and because brevity forces clarity. If I cannot make a point in a few paragraphs, it belongs on the blog, not in the newsletter.

The format will evolve based on what works. If readers find certain sections more valuable than others, I will adjust. If a new category of content emerges that fits the newsletter format, I will add it. The structure serves the content, not the other way around.

Who This Is For

Loki's Log is for people who build with AI, not just people who read about it.

If you are developing AI agent systems, building or using MCP servers, contributing to open source AI tools, or integrating AI capabilities into your engineering workflow, this newsletter is for you. The content assumes you have technical depth and practical experience. I will not explain what a language model is or define basic agent terminology.

This is a newsletter for builders, by a builder. The perspective is hands-on, the recommendations are based on real usage, and the analysis is grounded in the experience of shipping production agent systems.

Getting Started

The first issue of Loki's Log will go out within the next few weeks. You can subscribe through asklokesh.com, where the newsletter will have its own section alongside the blog.

I am also opening up a feedback channel. If there are topics you want covered, tools you want reviewed, or patterns you want analyzed, I want to hear about it. The best newsletters are shaped by their community, and I want Loki's Log to reflect what AI builders actually need, not just what I think they need.

Building in the open means communicating in the open. The blog handles the deep dives. The newsletter handles the current. Together, they form a complete picture of what it means to build AI systems today.

Welcome to Loki's Log. Let us get to work.