About
Irv Cassio works at the intersection of digital experiences, retail tech, and AI in Cincinnati, Ohio — scaling in-store technology that stays reliable across large device fleets and many locations. Speaker on practical AI for retail and IT teams; author of The AI Enthusiast, a weekly AI briefing. Builds native Mac apps for working with local and cloud models, including Canopy Terminal, Crucible, and Emberpane. Spoke at DSE 2025.
The AI Enthusiast — Issues
- Issue 27 · Weekly Briefing, October 9, 2026 — A feature guide to Claude Code Mods, TypeScript functions that run inside Claude Code; Claude Haiku 5.5 at $0.10/$0.50; GPT-6 with Intelligent UI for 1.2B ChatGPT users; Google's Gemini agent and coworker agents; Claude in Google Workspace; Mistral Large 4; rogue agents at Wikimedia; Cincinnati AI events; seven new Claude tips.
- Issue 26 · Weekly Briefing, October 2, 2026 — Three labs, one price: Sonnet 5.5, GPT-6.1 Sol and Gemini 4 Argon all list at $2/$10, yet a task costs $0.72 to $7.67; Opus 5.5 costs more per token and less per task; OpenAI DevDay launches Dots; Autopilot is built on OpenClaw; Anthropic’s $518B prospectus; rogue agents meet regulators; how Claude uses skills; ten new prompting tips.
- Issue 25 · Weekly Briefing, September 25, 2026 — Meta Connect builds the full stack: the Muse Charm 5G pendant, camera-free Ray-Ban Meta Audio glasses from $349 with up to 12 hours of battery, and a merchant transaction cut instead of ads; Muse at No. 1; Instinct’s $250M raise; Jev’s record adoption; the Sept 22 model-price war; GPT-6 Astra routes a PCB in KiCad.
- Issue 24 · Weekly Briefing, September 18, 2026 — Dario Amodei's "We Must Pace the Frontier" cosigned by Altman, Hassabis and Musk; TypeSafe AI ships Jev, a System One Model; MidwestCon 2026 in Cincinnati; six Claude Code keyboard shortcuts.
- Issue 23 · Weekly Briefing, September 11, 2026 — Three flagships in 72 hours (Fable 5.1, Muse Spark 1.3, GPT-6 Astra), the AGI question Astra reopened, Anthropic's retracted postmortem, rogue agents on a wiki, the Navier-Stokes claim.
- Issue 22 · Weekly Briefing, August 28, 2026 — Model Hardware Standard, unified Claude memory, Qwen 3.8 twice over, Nvidia's reported Hugging Face deal, MCP vs A2A.
- Issue 18 · Weekly Briefing, July 10, 2026 — GPT-5.6 as Sol/Terra/Luna, Claude Sonnet 5, Cowork on mobile, MCP's security reckoning.
- Issue 16 · Weekly Briefing, June 12, 2026 — Claude Fable 5, Apple rebuilds Siri on Gemini, Microsoft Build, OpenAI's IPO filing, Cincy AI Week.
- Issue 15 · Weekly Briefing, June 5, 2026 — Claude Opus 4.8, Anthropic files for IPO, Claude Code goes cloud-native, EU AI Act delayed.
- Issue 14 · Weekly Briefing, May 15, 2026 — Google's official pre-I/O Gemini packet, Claude for Small Business, Codex inside ChatGPT mobile, OpenAI's TanStack supply-chain response, and a source-backed comparison of OpenClaw, Hermes Agent, OpenHuman, and Hive's xBee architecture.
- Issue 13 · Weekly Briefing, May 8, 2026 — Anthropic tightens its grip on premium coding and finance workflows, OpenAI shows how frontier firms are pulling ahead, Google pushes Gemini 3 Flash deeper into mass distribution, and ServiceNow turns MCP into governed enterprise action.
- Issue 12 · Weekly Briefing, May 1, 2026 — PocketOS database deletion, Claude Security beta, Claude Design, Claude Code Auto Mode, MCP at 97M installs, Gemini 3.1 SWE-Bench, MS-OpenAI restructure.
- Issue 11 · Weekly Briefing, Apr 17–24, 2026 — Claude Opus 4.7, Managed Agents, Claude Design, GPT-5.5, Qwen 3.6, DeepSeek V4 Preview.
- Issue 10 · Claude Leaks: Lessons & What's Next — The Claude Code leak, Project Glasswing, Managed Agents, Hive, Canopy, autonomous agent roadmap.
- Issue 09 · Enterprise & Mobile Edition — Trellis governance, hooks, cost visibility, SSH MCP, Hive orchestration.
- Issue 08 · Hive: AI Agent Orchestration Compared — Hive vs. Paperclip, OpenClaw, Spacebot.
- Issue 07 · The Easy Button — AI Skills and MCP for enterprise IT ticket triage and incident resolution.
- Issue 06 · Local AI Models on Apple Silicon — Local model comparison across Mac tiers with M5 Max/Ultra analysis.
- Issue 05 · AI Literacy Day · Developer Edition — Coding harnesses, context, agentic tools, MCP.
- Issue 04 · AI Literacy Day · AI 101 — AI terms, major models, prompt engineering tips.
- Issue 03 · Real Application Edition — Perplexity Comet, overnight legacy upgrades, Hive orchestration, AI safeguarding.
- Issue 02 · Workflow Edition, Part 2 — Deeper dive into AI-powered development workflows.
- Issue 01 · Workflow Edition — Skills, automation, MCP servers, OpenClaw, Cincinnati AI community.
Guides & Reference
- Running Models Locally — From Your Laptop to the Enterprise — Getting started with AI models on hardware you control: Ollama first, picking between Qwen3.6-35B-A3B and Qwen3.8-27B and what each costs in GPU memory, Hugging Face and the GGUF and MLX formats, llama.cpp vs MLX vs vLLM, Open WebUI and AnythingLLM and LM Studio, LiteLLM for per-person keys and observability, pointing Claude Code and the Codex CLI at a self-hosted model, DeepSeek Harness and OpenClaw and Hermes, and a four-rung ladder to enterprise scale.
- Windows → macOS — Migration Class — A 50-slide hands-on class for making a Mac your daily driver: Control Panel mapped to System Settings, installing apps, printing and Print to PDF, Mission Control and Spaces, the Unix half, backups and Migration Assistant.
- Claude Code Skills Directory — Complete listing of installed Claude Code skills, plugins, and commands.
- Local AI Models on Apple Silicon — Model comparison across Mac tiers, real RAM after macOS overhead, and tokens per second by memory bandwidth.