Build something with Claude you'll actually use.
Six live sessions, twice a week, with the same small group of coaches, consultants and founders.
Phil Rivard
Techstars alum, coached 350+ founders. Tinkering with how I work since long before Zapier and Notion.
This is for you if…
You keep meaning to figure it out.
New model every month, new tool every week. The browser tabs pile up. You tried it once, it kind of worked, and you went back to how you did things before. You just need someone to show you where to start.
You get things done in Claude, but you're not sure how.
It works, sort of. Every time you want to change something, you relearn it from scratch. Chat, Cowork, Code… you're still working out which is which.
Everyone else seems to have it figured out.
Learning it alone hasn't worked. You want a small group, a deadline, and someone in the room who has done it. You're not behind. You're stuck, and so are most of the people I work with.
I've spent thousands of hours tinkering with how I work.
I trained as a programmer-analyst, and I've been testing and tweaking how I work ever since, long before Zapier and Notion. I try most new tools when they come out, and my brain maps pretty fast how they can work together. I switched from ChatGPT to Claude for how it writes, right before Claude Code came out, and I've been a heavy Claude Code user ever since their beta.
That being said, I'm not stuck in one tool. I still very much enjoy testing and adopting what the other labs ship. Right now that's OpenAI's dots and xAI's Grok Bot, AI agents you can give real work to.
My day job is coaching. I've coached 350+ founders through programs like Techstars, Creative Destruction Lab and NEXT Canada, and I co-built the Next AI Montreal accelerator. The work is often the same: find a problem worth solving, then help people build the thing. Part of that is knowing when Claude is the answer, and when it isn't. That's part of what we'll do here.
Before that, I started Nukern, a VC-backed company that went through Techstars Austin in 2017. We made it far, but not far enough. I learned a ton!
An email assistant.
It reads my inbox every minute, sorts what matters and drafts replies in my voice, checking my calendar when someone asks for a time.
Month-end invoicing.
It lines up the month's coaching sessions from my meeting notes, recordings and calendar, and drafts the invoices. Billing stopped being a chore.
Coaching prep with the full picture.
Before a 1:1, it pulls my notetaker, our emails since and my own notes, so I get a full 360 since I last met that founder.
A second brain.
I capture an idea in one tap; Claude files it, tidies it and connects it across my many personal projects and clients.
A pitch-practice tool for founders.
A timer and a script manager to rehearse and tighten a pitch. It's free and it's live at pitchsmith.io.
A design studio made of AI agents.
One writes the page, one designs it, and nothing ships until a taste check passes.
A family announcer for our schedule.
A Raspberry Pi that calls out our daily routines in a natural voice. Built, and still waiting for its spot in the house.
A live card game I built with my son.
Two players connect over the internet and play each other in real time.
A Pokémon restock watcher.
It pings my phone the moment a sold-out collector set is back in stock. Not everything has to be work.
A mic scrambler for video calls.
It distorts my voice just enough that transcription bots can't follow, while people still understand me.
All of these were built with a small AI team I run: agents scout the work, build it, review it, and ship it. I direct and approve changes.
I'll show you
how to build your own.
Registration closesOct 19.
You pick the problem. We build the solution together.
Don't know what to build yet? That's where we'll start. Setup week will help you find one thing worth building. It runs on your own data (your inbox, your calendar, your files), so it starts from your reality, not a demo. It won't have every bell and whistle, but it will work, and you'll know how to improve it. Here's the kind of thing people build:
An inbox that drafts your replies.
In your voice, ready for you to edit and send.
Prep for every client call.
Your past notes and the latest context, pulled before you join.
A weekly recap that writes itself.
From your calendar, email and notes.
Proposals and follow-ups in minutes.
Drafted from your templates and your past work.
The small tool you wish existed.
For the task you still do by hand every week.
Over the 3-week cohort, you'll build one thing that saves you time every week, made for how you actually work.
Ask → Chat. Do → Cowork. Build → Code.
It mostly comes down to three tools. You don't need to master all of them. You need to know which one to open, and when.
Chat
You need an answer, a first draft, or someone to think out loud with. Chat talks it through with you.
Cowork
You need the work done on your own files and apps. Give it the goal, let it run, check what comes back. No code involved.
Claude Code
You want to build a tool or agent. Claude Code is part Chat, part Cowork: it thinks it through with you, then builds alongside you.
Already deep in Claude and want a real partner to apply it to a startup? Read about my cofounder experiment
A setup week, then three weeks live.
Hands-on from the first session. Every session ends with something you can use that week.
- 6 live sessions
- Mondays and Thursdays, 1:00 PM ET
- 60 min + an optional 30-min office hour
- Slack in between
Setup weekGet set upOct 19–23
Registration closes Monday, October 19. Slack opens. You install Claude and Claude Code, connect your own data, and pick your first build, a morning brief or a weekly recap. We sort out installs together.
Week 1FoundationsOct 26–29
-
MonOct 261:00 PM ET
Make Claude your own, and which tool when
Meet the group, tour Chat, Cowork and Claude Code, and set Claude up your way: settings, connected tools, instructions.
-
ThuOct 291:00 PM ET
Prototype in claude.ai
Build your first one by hand on your own data, tune it until it's right, and have Claude write it up as a spec.
Week 2BuildNov 2–5
-
MonNov 21:00 PM ET
Automate in Cowork
Hand the spec to Cowork and schedule it, so it runs without you.
-
ThuNov 51:00 PM ET
Claude Code build day
Choose what you want to build next, set up your environment, move your instructions over, and start building it in Claude Code. Learn the agents, skills and harness you need to build well.
Week 3FinishNov 9–12
-
MonNov 91:00 PM ET
Break it and let it run
Break it on purpose, fix it, and let it run on its own overnight. Learn how to QA a tool before you launch it, and what to ask your agents to check for you.
-
ThuNov 121:00 PM ET
Polish and get ready to show
Tidy it up, run a security pass with your agents, and get it ready to demo.
The finaleShow and tellMon Nov 16
-
MonNov 161:00 PM ET
Show what you built
Its own 1-hour session, after a weekend to finish. Everyone shows what they built and how it helps their week. Knowing you'll show it is what gets it built, and you leave with a few ideas borrowed from everyone else's. Just us, no pressure.
This is a first run and I run it like a startup: session content and details may change as new tools ship or as we learn what works. The goal doesn't: you leave with something that works, built on your own data.
What you walk away with.
Tools you actually use in your business, and the know-how to build the next one.
The thing you built.
Working, made for how you actually work, and yours to keep improving.
A daily Claude setup that knows you.
Your instructions, your connected tools, the files Claude reads, and the workflows for the things you do most. Built around your actual week.
Your first piece of automation, written by you.
A Claude Code skill you write yourself. Most people don't think they can write one until they do.
Nothing disappears after week 3.
Recordings, transcripts and the Slack community stay open. When you forget how something worked in week one, you'll find it in week eight.
$750 CAD for the first cohort. $2,250 CAD for the next one.
First cohort
Ten people. It's the first run, so it's small and priced to test the format with you. In return you get a smaller group, more of my time, and the only time this class runs at $750.
The first cohort is more than a pilot. It's a conversation. I come in with a curriculum and a direction, and I adjust in real time to what you actually need.
For $750, you get:
- A setup week before we start, so nothing blocks you on day one
- 6 live 60-minute sessions, twice a week, Mondays and Thursdays at 1:00 PM ET
- An optional 30-minute office hour after every session, to ask questions, get unstuck or scope your build
- A 1-hour show and tell on Monday, November 16
- Recordings and transcripts of every session
- Private Slack community access
- Completion certificate
- Alumni pricing on the next course
If you decide it's not for you by end of week 1, you get your money back. No questions.
The 10 spots go to the first 10 paid invoices.
What you'll need: A paid Anthropic Claude plan (US$20/month minimum, US$100/month recommended), a Mac or Windows computer, and zero coding experience.
Testimonials
Nothing yet. What were you expecting? It's the first cohort.
After cohort 1 you'll get real quotes, real builds, and the receipts.
Questions you might have.
Can someone like me actually do this?
Yes. It's built for people who use AI a bit and haven't built anything with it yet. If you can follow along on a screen share, you can do this. No coding background needed, and setup week is there so nothing blocks you on day one.
I don't know what I'd build. Is that a problem?
No, it's the most common starting point. Finding the problem worth solving is my day job as a coach. We work it out in setup week and in week 1's office hours, from where your time actually goes.
What will I actually have at the end?
One thing you built for your own business, that you keep using, a Claude setup that knows you and your context, and the know-how to build the next one. Plus every recording and the Slack.
Will everything I make sound like AI?
No. That's what the prototype session is for. You tune the output against what you expected, and give Claude your own writing as examples, until what goes out still sounds like you.
What's the time commitment for the cohort?
Two 60-minute sessions a week for 3 weeks: six hours minimum, plus an optional 30-minute office hour after each session. If you want to retain what you learn, budget 3 to 5 hours a week on your build too. The point is to get that time back, and more.
What if I miss a session?
Every session is recorded and transcribed, and Slack is open in between. That said, live is where most of the value is, so plan to be there.
Who else is in the room?
Coaches, consultants and founders. People invested in making Claude (and AI) their own. Ten people max, so everyone gets real time with me.
Why Claude and not (insert favorite LLM here)?
I use Claude every day for how it writes and for what Claude Code can do. I still test what the others ship, and when something changes how we'd do the work, it comes into class that week.
Can I get my money back if it's not for me?
If you decide it's not for you by the end of week 1, you get your money back. No questions. Well... actually, I might ask what didn't work for you so I can improve the material, but it's ok if you'd rather not answer.
Take one of the ten spots.
$750 CAD for the first cohort, $2,250 CAD for the next one. Registration closes October 19, and the people in the room help shape the curriculum.
Save my spotThe 10 spots go to the first 10 paid invoices.