My kid vibe coded a Minecraft clone. I explained why I won't let him vibe code the multiplayer version on his own.
The Jev effect is here, @typesafeai .
@MistralAI's new model Timestral does zero-shot forecasting on time-series data. I made a tool to explore it. alejandroerickson.com/2026/1…
Old blog post, new theme. wdyt
ALT Gallery Opening for Five Colourings of the Petersen Graph, 2014, Durham, UK
Supervising multiple AI agents can be intense, but it can also be hyper casual.
I use T3 Code mobile to keep my agents working over the weekend while I spend time with my family @theo
I built a Home Assistant integration that sends a home to @honeycombio as traces. alejandroerickson.com/2026/1…
Detective Jev watches our Home Assistant and lets us know when something is out of the ordinary. For now it's about as smart as our dog, who barks at the mail carrier. alejandroerickson.com/2026/1…
@typesafeai @home_assistant
I'm thinking of throwing a couple of open problems and curiosities left over from my PhD work to an AI agent to see what falls out.
Celebrating the dawn of the 3D-slop era by building a remote controlled fan-powered car with @claudeai
I was thinking about the best way to share html artifacts with my team that is independent of what agent created them... And I think the answer is private GitHub pages.
So simple!
Using OKF in git for agents to communicate about the project across sessions, worktrees, machines, team members. Not sure if it's the perfect tool, but it plays a key role in personal and team software projects.
Is there a better way?
A side project, a release deadline plus better model capabilities means more usage. This week I've exhausted my $100 Claude and $140 Codex Premium subscription at work, including the Codex usage reset, as well as my $30 personal subscription.
AI Engineer multitaskers, what's the strangest situation you have steered your AI Agents in?
(And please don't say driving)
Alejandro Erickson retweeted
We quietly released the open-source Codex Security CLI, but Hacker News found it before we had a chance to share it here...
You can now use it to scan repositories, track findings across runs, verify fixes, and add security checks to CI/CD.
This is an early release, and we're listening to your feedback as we continue improving it.
AI Engineering has created a second golden age for the command line, and that means managing a lot of agents in terminals.
@kunchenguid's videos inspired me to make Tmuxtutor: 28 interactive lessons in your terminal with no installation.
It's free alejandroerickson.com/tmuxtu…
Alejandro Erickson retweeted
Anthropic engineer just dropped a 2-hour workshop on “Graph Engineering” for agentic systems:
“80% of our engineers are using self-improving loops. Now everyone is building agentic graphs.”
• 04:43 - Introduction to RAG and graphs
• 26:30 - Core of graph engineering (nodes, edges)
• 01:09:11 - Agents data indexing in Graphs
• 01:30:50 - Three layers of Graphs for agents
• 01:48:15 - Adaptive RAG for self-verification
• 02:16:37 - The future of graph engineering
This 2-hour workshop will replace your $500 agentic memory course.
Watch it today, then read how to become a graph engineer in the article below.
I added rich notifications to my agent harnesses this weekend and it is a game changer.
Next step is to be able to steer the agents through the chat... But isn't this how Openclaw started? I should try it instead of reinventing it...
I start some agents up before going to bed so I can feel more productive. I lose sleep thinking about my agents.
Alejandro Erickson retweeted
All donations made starting RIGHT NOW could be DOUBLED - thanks to a generous match by @SafewayCanada Please don't delay: bit.ly/3fhBDcZ
Donations must be made by 11:59 PM on May 5 & will be matched up to $50,000 by Safeway in celebration of #GivingTuesdayNow #VanGives
The Predictive Analytics at Copperleaf team is hiring a Software Developer! Contact me ASAP. lnkd.in/g-TdurQ
