Ship one workflow

Three days. One workflow. In production.

You don't need another AI tutorial. You need one thing working.

In three focused sessions, you'll choose a high-leverage workflow, map it, build it, ship it, and iterate on it.

Who this is for

This sprint is for founders, operators, and anyone responsible for getting things done who:

Keep hearing "you should automate that" but haven't done it. Have too many half-formed ideas and not enough finished ones. Are tired of manual, repetitive work. Don't need inspiration, need structure. Want applied help, not a tool tour.

What you'll leave with

By the end of Day 3, you will have:

One clearly defined workflow. A working AI-assisted version. Defined success metrics. A one-page automation brief you can reuse internally.

You won't leave with a mess of 20 potential prompts.

You'll leave with a system that's running.

Format

Three two-hour live virtual sessions. Small class size. Pre-work and homework required. Shared Slack channel for the duration of the sprint: ask questions, share progress, get unstuck between sessions.

$300 per participant.

What makes this different

Most people leave AI workshops with ideas. You'll leave with something running in your actual environment, timed, documented, and ready to hand off.

Ready to ship something real?

The next sprint is happening in March.

Apply for the next cohort

How it works

Session 1: Find the right thing

We talk about what's worth automating.

In the session, you’ll inventory real frustrations, map one candidate workflow, and define what “better” actually means in measurable terms.

You’ll leave with:

• One clearly scoped workflow
• A mapped current state
• A defined success metric

Homework (30–45 minutes):
Refine your workflow map, gather 2–3 real examples or artifacts you’ll use for testing, and clarify your success metric. Session 2 moves quickly. Showing up prepared makes it productive.

Session 2: Build the AI layer

We design and test.

You’ll learn how to structure an AI-assisted workflow: what it knows, what it's allowed to do (and what it shouldn't do), and what a good output actually looks like. I’ll walk through one complete example, then we move into working time.

This is a build session. You’ll draft, test, refine, and debug in real time.

By the end of the session, you’ll have a working version one.

(Working =/= perfect)

Homework (45–60 minutes):
Run your workflow. Time it. Note where it breaks, where you still had to heavily edit, where it hallucinated, and whether it actually saved time. Bring one example of a “bad” output to Session 3.

Day 3: Stress test and ship

Let's make this durable.

We’ll test your workflow against distorted inputs and edge cases, identify where it breaks, and define where human oversight belongs and refine it based on what you observe.

Then you’ll document it so it can run again without you reinventing it.

You’ll leave with:

• A tested workflow running in your environmen
• Clear boundaries and review point
• A one-page automation brief

There’s no additional homework after this session. The goal is that you leave with something live, documented, and ready to reuse.

Apply for the next cohort

Frequently Asked Questions

What if I don't know what workflow to choose? That's normal. Day 1 is built for that. Come with a list of frustrations and we'll narrow it down.

What tools do I need? Nothing exotic. If you can access ChatGPT, Claude, Gemini, or similar and your current workflow tools, you'll be fine.

What if my workflow is complex? We'll scope it down. Version one should be small and repeatable. 

Who am I?

I'm a consultant who has helped build AI-assisted content and operational systems for lean teams. Over the past year I've built production prompt infrastructure for a B2B data startup - a modular, multi-role system that turned raw podcast transcripts into fully distributed content across six formats and a 14-day posting calendar, contributed to a GitHub-based content intelligence system with 10+ specialized skill files covering the full content funnel, and integrated workflows connecting live data sources so a non-developer could run the whole thing without engineering support.

I help operationalize and design systems that hold up when you're not watching them.

This sprint is the applied version of that.