Eh-I
Stagione 1
The Wellness Check Your AI Agents Actually Need
Why Most AI Agents Are Crap (And What I Built Instead)
I Hired an AI Coworker That Lives in My Slack
Why Your AI Sounds Robotic (And How to Fix It)
di Shaun Whynacht
You built an AI agent to follow up with clients, and now it quietly runs in the background. But what happens the day the machine reboots, the internet drops, and your agent just stops, and you have no idea? In this episode, Shaun walks through the same problem he hit building an executive assistant agent on a local Mac Mini. The fix is a heartbeat: a small outside service that pings your system on a set cadence and pokes you when it goes quiet. He shares the free tool he landed on, healthchecks.io, how he set a 10-minute check with a 15-minute grace window, and why the monitor has to live outside the machine it's watching. Think of it like a wellness check for the family member who lives alone.
You bought an AI agent, tried it for a week, and quietly went back to doing things the old way. You're not alone. In this episode, Shaun explains why most out-of-the-box AI agents don't actually do what you need. They ask you to change how you work instead of working the way you already do. Shaun walks through the custom agents he built for Blue Cow Marketing over about two and a half weeks, starting small with one task: his meeting notes. He shares how that grew into a task system, drafted client emails, a sales dashboard, and a one-minute audio briefing every morning. He also gives you a simple first step you can do today. Fair warning: he says "crap" in the first ten seconds.
Try VIKTOR HERE You've tried AI tools that give you advice and ideas, but still leave the actual work sitting on your plate. This week Shaun talks about Victor, an AI he describes less like a tool and more like a coworker. Victor lives inside Slack, connects to the apps you already use, and does real jobs: building out Facebook and LinkedIn ad campaigns, generating ad creative, even watching a client's Loom video and making the edits. Shaun walks through four days of using it on actual client work, what surprised him, and how the pricing shakes out. No hype, no doom, just an honest look at where this is headed. Fair warning, you might start saying Victor's name around the house too.
Your AI keeps writing in a voice that isn't yours. You ask ChatGPT, Gemini or Claude for a newsletter, and what comes back sounds like a robot that read your website once. The fix isn't a better prompt. It's training. In this first episode of the EH-I Podcast, Shaun walks through the three moves he uses for his own accounts and his clients: how to "go on a date with your AI" so it actually learns who you are, how to use the deep research function to build a baseline knowledge file, and how to feed your own writing back in so it learns your voice. By the end, you'll be about 80% of the way to AI content that sounds like you. The "EH" is short for Canadian, by the way.
AI is moving fast, and most people only ever use the chat box. But your AI account has more going on than that, and knowing what each part does changes how much work it can take off your plate. In episode two, Shaun walks through the real sections you are seeing show up now: Chat, Projects, Cowork, Claude Code, and the Chrome browser extension that quietly does more than almost anyone realizes. He shares a real client story too, where 39 archived podcast episodes became two streams of social content and a 60 page book written in the client's own voice. The thread running through all of it is simple: feed your AI authentic content and use it to speed up the work you already do. Shaun keeps it practical, Canadian, and refreshingly free of hype.