Why hire me instead of an AI subscription?
An AI subscription writes code. I take the whole software side off your plate: I decide what to build and what to leave out, set the rules the AI works to, hire and lead the team when you need one, run it in production and answer for the result.
When a subscription is enough
Often it is, and then you don't need me yet.
- You want a clickable prototype to show a few potential customers.
- It is a landing page, an internal tool or a script that nobody's money or data depends on.
- You are an engineer with the time to set the rules and check the agents' work yourself.
- You are still finding out whether anyone wants the product.
Come to me when real users, real money or real data depend on it.
What I add on top
- Yesdoes it
- Partialonly part of it
- Nodoesn't
| Capability | Me, with AI agents | An AI subscription on its own |
|---|---|---|
| Writes code fast | Yes | Yes |
| Decides what to build, and what to leave outIt builds what you ask for. | Yes | Partial |
| Works to written rules and standards, reviewed every weekA subscription follows only the rules you write for it. | Yes | Partial |
| Sets up servers, deploys and backups, and keeps them running | Yes | No |
| Someone answers for the resultWeekly working software, free bug fixes for 30 days. | Yes | No |
| Costs less | No | Yes |
- Method
- Checked on November 9, 2026. Based on how I work today, described in my public agent rules (opens in a new tab), and on my own measurements in the two posts linked here.
- Fit
- If the subscription column covers what you need, use it.
Will you just run Claude and bill me for it?
Partly, yes. AI agents type most of my code, and I say so openly. You pay for the result, not for my typing: on a fixed scope the price is fixed, and code, accounts, servers and keys are in your name from day one.
- A written spec before any code.
- AI agents write the change, to my rules, style guides and standards.
- A separate reviewer agent checks it; sign-in, payments and secrets get a security review.
- Nothing reaches you until it passes that review.
- Every week I review the code and the agents' sessions: what slowed them down, what it cost and where money was wasted. Then I fix the causes, such as slow test runs or tests that fail at random.
- My own tool, Ringmaster, keeps the agents on task and their cost under control.
- You see working software every week, with a short written update.
The tools are not the hard part. I have vibe-coded too: without clear rules and a definition of quality, the result is mediocre and hard to maintain. Mine are public in spy4x/dotfiles (opens in a new tab), so you can read exactly how I work.
- pull requests merged in my first 30 days on Claude Code. Volume is not value, and much of the jump is not the model alone. Read the post
- 1,845
- how often the cheaper and the stronger model passed their first review. Review is where a person still earns their fee. Read the post
- 13% vs 33%
Already built it with AI?
Good. Keep going if it works. When real users, payments or personal data arrive, a second pair of eyes before launch is worth it.
- Code & security audit · From $1,500I check what is solid, what is risky and what to fix first.
- Strategy session · $150One hour to talk through your next technical decision.
Shipped to production
An AI tool can write code. It can't point to products it shipped, saw acquired, or still runs.
SmartLiteLiveFull-stack (web app + backend + infrastructure) · 2024–now
Built solo in 3 months; about 200 lamp poles in production since 2024.
FoodRazorOutcomeTech Lead · 2018–2019
Scaled across 10 countries and hundreds of restaurants; acquired in 2023
Acquired by OrderEZ, reported by e27 in February 2023. Source
CorecircleOutcomeTech Lead · 2021
Scaled to 200K+ users; acquired in 2024
The founder said 200K users in her June 2022 interview. Source
Questions
Can't I just prompt an agent myself?
You can, and for a prototype you should. What takes years is knowing what to ask for, spotting the change that looks right but isn't, and keeping it running once people depend on it. That is the part you hire me for.
Who pays for the AI tools?
I do. My AI subscription is part of how I work, not a line on your invoice. If you would rather it ran on your company's own AI account, for example to keep your code under your own contract, I can work through that one.
If you bill hourly, do I pay while the AI works?
No. I bill only the hours I actually work on your project: setting the agents' tasks, checking their progress and reports, judging the quality, writing new rules, and calls. Once I hand the agents a batch of work, the timer stops, and it starts again only when they have finished and I am back on your project.
I already built something with AI. Can you take it from here?
Yes. I start with a code audit, so you know what is solid, what is risky and what to fix before launch. Then you decide whether I continue, your team does, or you keep going on your own.
What if it turns out I didn't need you?
If in the first five days you feel this is not working, tell me and I refund what you paid. We start with one or two weeks of work. If either of us wants to stop at the end of it, we stop — you keep everything built so far.
Not sure which you need?
You own everything from day one
Code, accounts, servers and keys are in your name.
Free bug fixes for 30 days
Bugs in what I delivered are fixed free for 30 days.
Tell me what you are building on a free 30-minute intro call.