I've been preaching for a long time that the best way to do deals is only to do scoped projects: I was wrong. Since I've split Ghostcoded into training/coaching and then actual building and spinning up a new business called Sublevel with another technical founder, my eyes have been opened to the ideal work you can do between you and your clients. I was always worried about hourly work and how it would tie me down, but when i got a co-founder, they were able to take off that load. Now as their time is being taken up, we can get a plan together to begin the process of hiring contractors to be part of our team to fulfill on the work we are bringing in. So what i've found is that there are 2 paths your clients can fall in: Path 1 - the scope is crystal clear and well defined If they know what they are talking about and can explain it super well, then this is a good candidate for you, bc you can whip it out and not worry about the number of hours you put into it and kinda bllitz it. They are happy bc its predictable and a guaranteed outcome Path 2 - the dynamic capacity allows to ambiguous projects to be maintained and still help with expectations and budgets This path allows the client to "reserve" a given number of hours from you a month. as they secure more hours or longer monthly commitments, they achieve higher discount brackets. These are for the difficult and intricate projects whose scopes are very hard to define, you have alot of questions, they arent even sure all that would be included etc. These kind of projects are very dangerous bc scope creep and additional costs for the client can make a project go bumpy. how's everyone doing on the scoping game? finding clients? closing deals? anyone got any questions? I'll be online for a bit to help with anything

Reflecting on your journey to making your first significant revenue milestone, what was the one strategy or tool that flipped the switch for you? Share a short story or insight that others can learn from!

So I switched everything to local models. I swapped the frontier model for a Courier model from Jackson Oaks and now I’m running all the inference I want 24/7 with zero token costs. A lot of people ask how Courier actually works because it feels almost too good to be true. Here’s the real progression most people go through: 1. Start on the free tier just to test it 2. Move to Pro at $25/mo 3. Max out the monthly tokens and jump to the $100/mo tier 4. Hit the limit and choose your path: 4a. Buy your own Mac hardware and run Courier locally for unlimited tokens under your control 4b. Stay on Courier cloud and work with a verified provider (like me) for production-grade inference + a custom token limit Either way you’re on flat-rate AI usage instead of watching token costs climb every month. Token prices keep going up and a lot of the VC-backed plays are going to feel that pressure. If you own the inference layer, you don’t have to care. I’m all-in on this approach and honestly more excited about it than anything else right now. Anyone else here dealing with painful token bills on agents or other workflows? Have you looked at Courier yet? ghostcoded.com/courier

There are many misconceptions about starting and running an AI/automation/dev agency. What is one myth you believed before starting, and how did reality differ? Drop your thoughts below and let's debunk these myths together!