Husband. Dad. Navy Vet. Former AI Teacher & Tech Coach. Now I teach practical AI and build things through https://xitter.cf/t.co/U10t0oSQVt, Source4AI, and MealReels! ✝️

Florida, USA
Joined April 2022
Local AI deserves a flight recorder. Spark Overseer for @MiaAI_lab’s sparkDash: instrumented tool calls, command outcomes and local review signals on DGX Spark / GB10. No prompt logging. No AI reviewer tokens. Observe-only, not a firewall. PR: github.com/MiaAI-Lab/sparkDa…
1
1
1
242
Submitting a PR for @MiaAI_lab’s awesome SparkDash to oversee behaviors for those worried about this 👍
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected. Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions. Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent. But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models? We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem. Read the full report: cohere.com/blog/concerning-b…
6
1
21
5,388
Local AI deserves a flight recorder. Spark Overseer for @MiaAI_lab’s sparkDash: instrumented tool calls, command outcomes and local review signals on DGX Spark / GB10. No prompt logging. No AI reviewer tokens. Observe-only, not a firewall. PR: github.com/MiaAI-Lab/sparkDa…
1
119
Take back your TV!
OmarchyTV is officially live on GitHub! 🚀🖥️ Huge thank you to @dhh and @kdaigle for helping get my GitHub account back online so quickly. 26 MIT themes, AI TUI, MCP, Android TV, webOS & PC plugin: github.com/piermark/omarchyt… Computers and TVs should be fun and look cool.
1
166
Steve Darlow retweeted
So uh...I guess it's time to announce tstack? I caved. I am publishing my skills now.
199
64
99
2,453
387,339
Steve Darlow retweeted
That's what I'm talking about! Keep it coming. The long-term goal for sparkDash is not just a monitoring tool for your models. If you have great ideas, pls DM, open PRs, whatever!
Submitting a PR for @MiaAI_lab’s awesome SparkDash to oversee behaviors for those worried about this 👍
8
2
55
5,453
Steve Darlow retweeted
The next version of my New York digital twin, Valdrada 0.6 is out! Three.js or UE5 as renderers, comes with prepackaged clients for CARLA-like simulation or data collection or digital twin workloads. If you are unimpressed by low quality Opus city slop, this is the project!
13
12
1
187
9,542
Telegram is down! 😩
2
376
Steve Darlow retweeted
🚀 RedGPU V4.4.0-Alpha is out! A major WebGPU update bringing large-scale open-world rendering to the browser: GPU-driven Terrain (Landscape), Foliage & Grass, Interactive Water (Lake), Global Wind, and KTX2 Compressed Textures. ✨ What's New in V4.4.0-Alpha: 1️⃣ GPU-Driven Terrain (RedGPU.Landscape) • Async 16-bit heightmap tile streaming & global heightmap fallback • Continuous LOD geomorphing & GPU frustum/occlusion culling • 4-channel multi-layer PBR splatting, stochastic tiling & virtual texture atlas baking 👉 Terrain Demos: redcamel.github.io/RedGPU/ex… 2️⃣ Large-Scale Foliage & Grass (Foliage & Grass) • GPU procedural scattering driven by terrain layer weights & slope • 3-stage mesh LODs + runtime-baked Octahedral Impostors with dithered crossfade • Chunk-streamed procedural grass with terrain ground-color blending 👉 Open World Demo: redcamel.github.io/RedGPU/ex… 3️⃣ Interactive Water System (RedGPU.Water.Lake) • Depth-based optical absorption & soft shoreline fade • Screen-Space Reflections (SSR), dual-normal refraction & underwater solar caustics • Real-time GPU 2D wave ripple simulation reacting to moving characters & objects 👉 Lake Demos: redcamel.github.io/RedGPU/ex… 4️⃣ Global Wind System (scene.windManager) • Unified wind direction, speed & strength synchronizing tree foliage sway and lake waves in real time 5️⃣ KTX2 Compressed Textures in BitmapTexture • Load .ktx2 textures directly via RedGPU.Resource.BitmapTexture • Native KTX2/ZSTD parsing & automatic GPU format negotiation (BC7, BC6H, ASTC, ETC2) • On-demand Three.js Basis Universal WASM transcoder integration for a lightweight core bundle 👉 KTX2 Demo: redcamel.github.io/RedGPU/ex… 🔗 Explore RedGPU: • Live Examples: redcamel.github.io/RedGPU/ex… • Documentation: redcamel.github.io/RedGPU/ma… • GitHub: github.com/redcamel/RedGPU #RedGPU #WebGPU #JavaScript #TypeScript #3DGraphics #GameDev #OpenWorld #WebDev
7
29
3
246
11,781
Steve Darlow retweeted
introducing tryboxi.xyz your agents get their own computer. on your mac. > disposable microVMs for Claude Code, Codex & Cursor > ready in 20 – 50ms > secrets never enter the box > runs 100% locally and it's open source 👀
45
20
9
356
21,453
Steve Darlow retweeted
Currently chatting with my Hermes Agent. I just gave it a Github repo with code on how to get access to this M5 StackChan. I told it to use both Astra and Opus to improve the code and gain access with a usable UI and phone app. It's been going for awhile now. I'm stoked
1
1
1
153
Steve Darlow retweeted
Alright, Bdots are now called Coworkers bdots was a terrible name github.com/fayazara/coworker… This is a very experimental project, I am still figuring out a ton of things as I build this (hence the delay). This is not an official launch. Here's a video of a bot that sends me a digest from multiple new sources I read, it sends me a message every morning at 7am and also I also asked it to create an app for me using Cloudflare dynamic workers (I feel this is the perfect use case for dynamic workers) More on how coworkers work A bot can be waked up by 1. You message 2. A routine (DO Alarms) 3. You email the bot [email protected] (CF email routing) 4. A webhook 5. Another bot 6. Someone edits a shared app. Each bot has rules so they dont go wild - Edit files - Run commands - Read the web - Drive the browser - Use connected apps (MCP servers) - Message other bots - Schedule routines A bot can - chat - Use Computer (cloudflare/computer) - Read the web (Kitesurf) - Drive a browser, take over. - Follow rules, Check in, Create routines. - has isolate memory or if you explicitly mention - global memory. - message other bots - create apps - can learn from you by looking at your browsing session (very rough still) Architecture - Each bot keeps everything in its own SQLite. There's no separate database. - You sign in to an app once, on the hub. Every bot uses that connection. - The worker only lets you reach your own hub and your own bots. AI Gateway - BYOK or use CF AIG unified billing. - My demo is ugin - deepseek-v4.1-flash-1017 (its really good and surprisingly cheap)
Okay, got way too many DMs, its already open and available. I am working on a little self deploy app, should be up soon
17
8
2
178
12,406
Steve Darlow retweeted
emblem is now open source a ~1:1 replica of @monogram_ai built with claude code & codex $6.4K worth of tokens at API prices, all through my subscriptions
19
5
1
119
8,417
Steve Darlow retweeted
Our small business (an MEP firm) built an AI harness for our engineers. Now it's good enough that we're using cloud agents to improve the harness itself. We created Mutiny BIM inside that platform because Autodesk couldn't support our goal: 100% vibe-drawn engineering plans by Q2 2027. We also plan to open-source the BIM portion. Imagine a building owner prompting permit drawings into existence, or a contractor requesting a ductwork revision and seeing it update in real time. Or getting their own revision sketches by riffing with their Grok Bot or Dot or Muse. Licensed engineers remain involved to validate designs, answer questions, and take responsibility for stamped drawings. No more waiting weeks for RFIs and drawing revisions. Genuinely hard to believe how fast this all is happening for our business @PermitZIP and what small companies are able to create for themselves now. #MutinyBIM
19
10
3
78
3,530
Steve Darlow retweeted
Your DGX Spark can now edit photos and turn stills into video on the same box that runs the full LocalRouter stack. Say "make it night", and the edited picture comes back in about 12 s. Give it a still, and you get up to 8 s of 1280x704 video with sound. This round was about some bug fixes and updates that I realised were missed. - Image editing is now in - Larger video generation (up to 8 seconds @ 1280x704p) - Smaller voice quant with the same quality by @plotarmordev Early ALPHA features coming soon! - Headless with companion. AI on the Spark, UI on Mac via Tailscale. - Decision model with computer use. - Update to the latest TensorFold engine by @ashxhart
2
4
1
66
3,667
Steve Darlow retweeted
TensorFold 1.1.4 just landed, and the voice is finally on time. Most of my 13 PRs are about the voice. With one optional setting, it now starts talking in under a second on a DGX Spark, and it can redo a spoken reply that breaks halfway. Sound on 🔊 github.com/ashhart/TensorFol…
9
6
1
53
3,390
Steve Darlow retweeted
👇 👇 👇 ::: new prototype drop:::: Keys - Ghost Machine (Audio/Visual agent): Almost instant speech to speech interaction with your agent powered by Qwen Audio Agent Concept and my man @Tech2Wild tony’s eyes of Kai interactive visual reference machine. Integrated into a (audio/visual interactive agent) Runs super fast , Powered by TensorFold technology to enable real time interactive speech to speech interaction that can drive your speech input into Hermes and all the agentic work flow you want to get done without ever typing a word securely remotely on your phone away from the desk or at your desk. …or on a CyboPal once it is released (to be tested). Repo in thread, try it out let me know. :
7
5
18
1,437
Steve Darlow retweeted
i just re created kurzgesagt's entire channel in 10 minutes with claude opus they have 0 competition because every video takes a full team of animators and 1200+ hours to make i turned it into a claude prompt that im giving out for free open claude code, paste the prompt, pick an idea, and your video is fully automated thank me later ↓ <inputs> Ask me for: one topic as a "what if" question (default: "What if the Sun vanished?"), the length (default 30 seconds), and whether I have an ElevenLabs key (if not, use a free local voice). </inputs> <direction> A Kurzgesagt-style animated explainer at 1920x1080, 30fps, drawn entirely in code: one 2D canvas plus one WebGL canvas for planets. Space: a vertical gradient #081d2c to #102348, five huge parallax nebula blobs (teal rgb 24,150,160 at 20%, blue 40,95,195 at 16%, coral 220,95,110 at 10%, indigo 60,70,175 at 15%), three star layers at parallax depths .12/.35/.6 that twinkle, 40% of the near stars drawn as 4-point sparkles. The Sun: radius 300px, half off the left edge. A radial gradient #fff7dc to #ffc857 to #ff8f4a with an offset highlight, additive halo rings at 1.23/1.48/1.76/2.1x the radius, 18 faint rotating ray wedges, 28 drifting glowing granules on its surface. Earth: an orthographic sphere in a fragment shader, 4x supersampled, from a 2048x1024 map painted from longitude/latitude polygons: land #5fc47a, forest #48ad6b, desert #f5c75f, rainforest #3a9d62, ice #f1f8ff, ocean gradient #4dc7eb to #1a66b8, white capsule clouds. Tilt .26 rad, spin .55 degrees per second. Day/night uses smoothstep(-.12, .24, N·L) with the light direction recomputed from Sun to Earth every frame. Navy night side with ~380 warm city lights, a coral band on the terminator, a cyan rim, ocean specular, an atmosphere halo that is stronger on the lit side. A small astronaut mascot (~180px) built from rounded rectangles: suit #f3f7fb, orange #ff8f4a boots, a visor gradient #1f5a73 to #0a1c2c with glowing cyan eyes that look at whatever matters, a blinking antenna light, an idle bob, squash and stretch on every pose. Type: Fredoka 600/700 only. Captions in caps with 3-6px letter spacing in teal #8fd8e6, gold #ffd98a or white, sitting in rounded pills of rgb(6,20,40) at 62%. Every line of text rises 1.3x its own height through a clip on an ease-out-quint. Finish: a vignette to rgba(4,4,20,.45) and a 6% film grain re-randomized 24 times per second. Banned: outlines, flares or loops on the Sun's edge, static infographic strips, full stops in on-screen text, text that fades in whole, AI images or AI video, Kurzgesagt's own characters, bird or music. </direction> <structure> 30 seconds, every line of narration gets a camera move or a transformation, nothing ever holds still. Tight on the Sun at 2.25x (0s) → on the word "vanished" it swells 6% for 0.22s, collapses in 0.33s, then a white flash and a shockwave ring (1s) → pull back to 1x revealing Earth still lit and the astronaut staring at the empty spot (1.6-3s) → on "eight" an 8:20 countdown slams in from 1.7x to 1x and starts ticking, a "LAST LIGHT" label rides the edge of the growing dark disc (3.2s) → dive to 2.3x on Earth with light streaks flowing in (5.6s) → on "you" a YOU pin drops (7s) → on "left" a photon lands on the pin with "THIS LIGHT LEFT THE SUN / 8 MIN 20 SEC AGO" (7.5s) → on "play" the pin morphs into a play button, "YOU PRESSED PLAY / 15 SEC AGO" (8.8s) → one pull-back from 2.4x to a 0.3x orbit view, "LAST LIGHT" and "GRAVITY" riding the ring (9.6-12.5s) → on "nothing" a dotted circle at the empty centre (13.4s) → on "hits" the darkness sweeps Earth in 0.7s, city lights pop on, the countdown turns gold at 0:00 (15.8s) → on "black" the background fades almost to black (17.4s) → Jupiter appears, "STILL SHINING FOR 1 HOUR+" (17.9s) → zoom out, a dashed tangent line, "30 KM/S" as Earth flies straight off (20.8s) → a thermometer counts 15 to -70°C while frost grows from the poles (22.8s) → a 0.8s rewind on a separate story clock (24.3s) → sunglasses drop onto the astronaut (25.2s) → a photon tagged "8:20 OLD", then the title (27.7s). For any other topic, keep the same shape: the event happens on its word, pull back to show scale, dive in to make it personal ("you"), pull back out, the consequence with a live counter, an escalation, a funny beat with the mascot, a callback and the title. </structure> <build> 1. One HTML file. Every frame is a pure function of time: window.seek(t) draws frame t. No timers, no CSS animation. 2. Camera moves on easeInOutCubic, zoom is interpolated in log space, every pop is a damped spring (damping .42-.5, frequency 12-16). 3. Voice: ElevenLabs eleven_v4 with a designed voice (calm, warm, curious male, late 30s, crisp, light neutral European English), stability .5, similarity .75, generated with timestamps. Without a key, use kokoro-onnx locally and get word times with faster-whisper. Every animation event fires on its word. About 165 words per minute, 0.24s between sentences, a 0.36s beat before the big reveal. 4. The darkness front: any pixel within the front's radius from the Sun turns to night with a 30px soft edge. The front moves at ~47px/s and ramps to 700px/s near Earth. 5. Keep two clocks: ambient time for twinkle, spin and bob, and a separate story clock for events, so a rewind stays consistent. 6. Render with Playwright in headless Edge with the GPU: seek every frame at 30fps and blend 2 subframes (shutter 0.6) for motion blur, then encode with libx264 at CRF 16 for the master and CRF 20 for delivery. 7. Sound: a calm royalty-free space music bed 7dB under the voice, timed so its natural ending lands on the title, swelling 5dB under the title. One downloaded royalty-free sound effect per event kind (collapse, impact, pull-back whoosh, countdown tick, dive, pin tap, photon land, play click, light hit, reveal, frost, rewind), each peak-aligned to its frame and 4-10dB under the voice. Master with gain plus a limiter to -14 LUFS, -2.2dB true peak. </build> <gotchas> Make the WebGL globe canvas at least 2 x radius x zoom x 1.18 pixels or the sphere clips flat in close-ups. Never draw prominences on the Sun's edge, they read as white hooks. Never pad the voice to fit the visuals, tighten the read instead. Speed the darkness front up near Earth or it takes 13 seconds to cross the planet. Show me a contact sheet of every beat before rendering. </gotchas> <start> Ask me the inputs, write the narration with the beat each word triggers, show it to me, then build and render the video. </start>
kurzgesagt has 25M subs and they still go viral with 15M+ views on every single video for years nobody copies them because every video takes a full studio of animators and months of work i rebuilt their format with opus 5.5, 100% code for $0 per video, on the left is my own version i created he probably averages $150k+ earning per video 9 years ago he posted how he makes 1 video in 1200 hours, i re created this in literally 10 minutes for free (check in comments) would u guys be interested if i turned this into a free claude skill you can use?
175
186
52
3,007
230,254
Steve Darlow retweeted
Education doesn’t have to be text and flat diagrams. So I’m building a 3D book about energy that students can play with: 10 lessons to predict, test and prove. Built with Opus 5.5 and Three.js. Try Lesson 1, Energy stores: everyday objects and where their energy is kept ↓
95
290
37
2,280
165,214
Steve Darlow retweeted
img2character is now available as a separate plugin! I've finished splitting the character workflow out of img2threejs. You can now try it yourself, including character rigging and animation. I'm also working on a more optimized upgrade, which I plan to release next week. More to come! Give it a try and let me know what you think. 🔗 github.com/img2threejs/plugi… Install: img2 add img2threejs/plugin-character
5
14
1
96
5,850