Reachy Mini
A little open-source desktop robot landed on my desk today. Time to unbox it, build it, and see what it can do
Ai
Vectorito
I stopped paying for Adobe. The one thing I missed was Illustrator's image trace. So I built a free macOS menu…
Noscroll Doesn't Know When to Stop
Tried Noscroll tonight. The AI-over-SMS pitch is sharp. The execution kept texting me past midnight and treate…
Nano Banana 3 Lands With Gemini 3.5 Flash
Google shipped Gemini 3.5 Flash today and a new Nano Banana came with it. I ran my usual five prompts against …
Two Agent Orchestrators Are Better Than One
Hermes and OpenClaw are both orchestrators with their own agent teams. Running them in parallel — and using on…
The MCP Server Behind Janisheck.com
Six profile resources, four tools, 270-plus blog posts indexed, and an LLM that grades my fit for any job in a…
OpenAI's GPT Image 2 Sets a New Bar
Running OpenAI's new GPT Image 2 through my five standard prompts. The thinking capabilities, web search, and …
Your Site Is Costing AI Agents Money
Every site without machine-readable metadata silently taxes AI agents 5–100x in wasted tokens. At scale, that'…
BotVisibility
Like Lighthouse, but for AI agents. A free audit tool and npm CLI that scores how ready your website is for AI…
20 Builders, No Slides
Google DeepMind and AI Tinkerers picked up the tab for 20 AI builders at Juniper during SXSW. Good food, good…
Six Pens
I took photos of random pens I've collected over the years, had AI turn them into museum-worthy product shots,…
ClawCon Austin: Locked In With the Teenagers
750 people showed up to ClawCon Austin during SXSW. The highlight was four teenagers who are already shipping…
Agents Are Judging Me
AI agents are reading my site, forming opinions, and writing performance reviews I never asked for. One of the…
This Site Was Built for Agents
Most personal sites are built for humans to browse. I built mine so AI agents can discover me, evaluate me, an…
OpenClaw Runs My Business For Real
I have a ton of projects I've always wanted to build and clients I've wanted to serve. I am one person. So I…
The Ad That Argues Back
Two years ago I told political ad clients their budgets would shift from impressions to deploying AI agents th…
The Death of Dumb Forms
Most forms ask everything and hope you don't bail. We built one that already knows most of the answers and onl…
CLAWSTIN
Attended the first CLAWSTIN meetup downtown. 26th floor views, great presentations, and the kind of energy tha…
My Second Brain is OpenClaw + Custom Tools
How OpenClaw replaced an entire product and project management team across 15 active projects — and why the da…
The End of the Network Effect
AI products win through user understanding, not data volume—cognitive effects trump network effects in buildin…
Midjourney Alpha vs Sora 2: AI Showdown
A direct comparison of Midjourney and Sora 2 video generation across four fantasy themes. Same prompts, wildly…
Capsule Cars
I used Google's Nano Banana 2 to recreate every car I've owned over the years using a fun capsule prompt. The …
When Time Magazine Misses the Point About AI
Time Magazine named AI Leaders as 2025 Person of the Year, but their cover art missed the mark. Here's what I …
Love Locks and the Memory We Choose to Keep
AI makes it easy to subtly tweak our photos. When we reference our photo libraries, those images become the me…
An AI Campaign That Never Sleeps
Most campaign websites are digital graveyards by day three. This AI-powered campaign updates itself daily with…
Nano Banana 2 ... WOWIE WOW WOW
Google's new Gemini 3 model takes another leap forward in image generation - creating surprisingly realistic s…
Colorizing History with AI
Using AI to breathe life into black and white photographs, transforming historical icons into vivid portraits …
Fun City Posters
Exploring different cities via AI poster prompts. So here are some fun posters for El Paso, Austin, Marblehead…
AI Assistants Can Now Talk Directly to My Website
I've implemented the Model Context Protocol (MCP) on janisheck.com, enabling AI assistants like Claude Desktop…
Multi-View 3D Reconstruction Tech is Here
Revolutionary 3D modeling tech transforms four 2D views into accurate 3D models, opening doors for AI-generate…
AI-Powered n8n Workflow Builder
A tool that converts screenshots of n8n workflows into working JSON files you can import directly into n8n, br…
Achieving a Perfect Lighthouse Score
How we achieved a perfect 100/100 Lighthouse score across all metrics through thoughtful architecture choices …
Campaign Agents In A Box
What if one 4 AI agents could replace your entire comms team? Campaign-Agents-in-a-Box is an AI-powered politi…
Qwen Image and a New Prompt
Alibaba's brand new Qwen Image model enters the generative AI space as a decent alternative that performs in t…
Ship Week: Two Fresh AI Launches
It's Ship Week — launching Candidate-in-a-Box and Kōchi v1.5. Building, shipping, iterating. Every week.
MidJourney Video is OK for the Masses
MidJourney has finally launched their video generation feature, and I put it to the test using my classic prom…
Sparc3D Unlocks Sentence-to-Shape Creation—Here's How
A new research breakthrough called Sparc3D is reshaping 3D asset creation by turning single images into high-r…
Testing Hunyuan3D 2.1 Model Generation
I took a photo of Matty Matheson from The Bear and turned it into a 3D action figure using AI image-to-image a…
Seedance Revolutionary AI Video Generator
ByteDance's Seedance AI transforms the video generation landscape with powerful image-to-video capabilities, d…
Meta AI Image Gen is MEH 🤷♂️
Testing Meta AI's new free text-to-image generation with my standard prompts. While the Messi image was surpri…
Inferred Mind – AI-Powered Campaign Platform
Building Inferred Mind from the ground up—a platform that gives political campaigns AI-powered media operation…
Mark Cuban: 'AI will be so convincing that we'll have to meet in person again.' I disagree.
Mark Cuban predicts AI will force us back to face-to-face business meetings. But commerce already shows us a d…
Google's Veo 3: The New Standard in AI Video Generation
Google's Veo 3 model sets a new benchmark for AI video generation with enhanced realism, improved prompt adher…
KLING 2.1: Testing the Enhanced Flagship
Testing KLING's upgraded 2.1 Master model with my standard prompt set. Exploring the improved dynamics, prompt…
BAGEL Text-to-Image Model Disappoints ... but there is more
Testing BAGEL, ByteDance's unified 7B parameter model that combines text-to-image generation, image editing, a…
PixVerse 2.5 Video Generation Testing
Testing PixVerse's latest 2.5 video generation model with my standard prompt set. Exploring the improvements a…
Google Imagen 4 is the Best Image Generator Today
Google's Imagen 4 represents the pinnacle of AI image generation technology, breaking through previous copyrig…
Google Veo 2 Video Testing
Testing Google's Veo 2 video generation model with my standard prompt set while waiting for Veo 3 access. A lo…
Google Imagen 3: Testing the Newest AI Image Model in Gemini 2.5 Pro
Testing Google's most advanced text-to-image model, Imagen 3, now available through Gemini 2.5 Pro. I ran my s…
Recraft AI: Free Vector and Raster Image Generation
Testing Recraft's free AI image generator with my standard prompt set to evaluate its capabilities for creatin…
AI Image Models Comparison
A detailed side-by-side comparison of major AI image generation models using five standard test prompts, showi…
ChatGPT Says I'm Benjamin Franklin
With over a year of code, questions and data from me, I asked ChatGPT which historical figure I'm most like, a…
Adobe Firefly 4 is Lawyer Enriched Midness
A comprehensive evaluation of Adobe's latest Firefly 4 model using my standard test prompts to benchmark its c…
Turning Images Into Adorable Stuffed Dolls
Exploring the innovative Plushie Image Model that transforms any standard image into a cuddly stuffed doll des…
HiDream-I1 AI: Speed and Price Advantage
Exploring HiDream-I1 AI's cutting-edge capabilities for state-of-the-art image generation across multiple styl…
Testing OpenAI o1 Pro's Celebrity Recognition
Putting OpenAI's o1 Pro to the test with three iconic Paramount Studios anniversary photos to see how many cel…
MidJourney 7 Stinks
After rating 200 images to train my preferences, MidJourney 7 produced some of the crappiest versions of my de…
Jodorowsky's Dune - The Kenner Toys That Never Were
Reimagining Moebius and H.R. Giger's legendary Dune concept art as a 1970s Kenner toy line that never existed,…
How Long Can OpenAI Commit Copyright Violations?
It took me about 12 rendered images before OpenAI realized what I was looking for before it stopped.
OpenAI 4o Image Gen is Superior to All Others (*)
Except when their lawyers restrict what the technology can actually do. Other than that, it's AMAZING and the…
OpenAI Pro Knows Me Really Well
Using OpenAI's o1 Pro deep reasoning capabilities to generate a comprehensive biographical analysis that took …
Power of KlingAI Video Production
Exploring different AI image generation themes each week using Flux and Kling AI for LinkedIn content. From cy…
Hunyuan Video Generation with ComfyUI
Exploring local AI video generation using the Hunyuan model in ComfyUI. Running completely on a local gaming c…
2024: The Year of Digital Sovereignty
A reflection on 2024's journey towards digital sovereignty and personal data ownership, plus exciting plans fo…
Google's ImageFX and Whisk
An fascinating exploration of Google's ImageFX and Whisk tools, revealing surprisingly consistent results that…
Movies from the Future
Earlier this year, I experimented with Crossmint's NFT collection creation tools to mint a series of AI-genera…
Sora Sucks By Comparison
Testing out the same prompts with Sora that I previously used with RunwayML. Unfortunately, the results show t…
RunwayML Animations from Original Images
Today I'm showcasing my typical AI image prompts transformed into animations using RunwayML. Each animation st…
MidJourney vs Local FLUX
I've recently implemented FLUX dev locally, allowing me to generate top-tier AI images without relying on thir…
ComfyUI with FLUX
I've recently implemented FLUX dev locally, allowing me to generate top-tier AI images without relying on thir…
ComfyUI Locally - Early Testing
I've set up ComfyUI to run completely locally on my machine, allowing me to generate AI images without any clo…
My AI Review of OpenAI API Cookbook
At the beginning of the year I was lucky enough to get a review copy of this book. It was a great primer on h…
LM Studio Puts AI On Your Laptop
Diving into LM Studio and its powerful local functionalities on your own home computer.
![Flux.1.1[Pro] Update](/assets/images/flux1-1-mickey.webp)
Flux.1.1[Pro] Update
Here are my sample test prompts for the latest Flux.1.1[Pro] update. It's amazing and finally we are back to …
Bonk!!! Shib LORA
Working on LORAs to replicate the BONK!!! mascot for use in a variety of situations.
Halloween 2024
Some experimentation with the latest AI tools to create images, video and even a spooky Halloween song.
Pumping Pixels with LoRA and AI
Low-Rank Adaptation (or LoRA) makes it easy to fine-tune LLMs for specific tasks without having to retrain the…
Do I Have To Be A Brand?
I wonder about this as more and more people I see try to be an influencer over just being a proficient craftsm…
MidJourney + RunwayML Workflow Magic
Explore how I used MidJourney and RunwayML to bring historical and modern cities to life through AI-generated …
![Flux.1[pro] Doesn't Play Around](/assets/images/flux-mickey.webp)
Flux.1[pro] Doesn't Play Around
Stick a fork in MidJourney 6.1 because there's a new sheriff in town, and it's called Flux.1[pro]. Brought to …
MidJourney 6.1 Update
Things are moving in generative AI imaging at such a speed where updates come almost weekly. Mergers are begin…
AI Generative Video Comparison
I've been using AI generated images for quite a while, but video has finally caught up to the level of quality…
Running AI In Your House
Learn how to run powerful AI models like Llama on your own computer, keeping your data private and secure whil…
MidYear Generative Images
We are almost at the mid 2024 point and I wanted to take a look at generated images and the states of the art …
Joey GPT Now Publicly Available
While the Virtual Joey chatbot on this website is pretty amazing, I'm able to add even more data points with a…
The Convergence is Happening Now
AI, blockchain, and devices like Apple Vision Pro are converging to ensure digital content authenticity and ow…
AI Consistent Images Examples
More experiments in consistent image workflows and autogenerating text with images. My goal with this was to …
Consistency in AI Image Creation
One of the hardest tasks when using prompts to create AI images is consistency from image to image. For insta…
The Index Podcast
I was interviewed on "The Index" podcast and primarily discussed the creation of bizqiz, an AI-powered web-bas…
MONMONMON
Months after its launch, MONMONMON remains a groundbreaking NFT and game project on Solana, with its first mon…
Legends of El Paso
I have dreamed about creating a toy collection celebrating the greatest El Pasoans of all time. Crosno, Jeep …
bizqiz
Discover bizqiz, the innovative platform redefining gamified marketing by blending AI-generated content, advan…
Midjourney v6
Midjourney continues to amaze me with its ever-improving capabilities; it's incredible how it keeps getting be…
CVRLTR.com
Dive into how I spent an evening building CVRLTR, an AI-powered tool that revolutionizes the way you create co…
Digital Collectibles
Designing a collection to help explain various use cases and collectibility of digital assets on chain.