Build notes · updated October 30, 2026
How I built Be in the Field: the stack, the costs, and who I brought in
I built Be in the Field, a native iPhone app, on my own with AI coding tools and a budget of about $300 a month. Codex wrote most of the Swift, Claude helped with design and iteration, the backend runs on Render, the marketing site on Vercel, and payments go through Apple's StoreKit. When the app was real and people were using it, I paid an engineer to refactor the foundation rather than rebuild it. Marketing is TikTok, Instagram and customer calls. There is no one right path from zero to one; this is the one I took, with the receipts.
By Lily Russo, founder of Ossur Labs and maker of Be in the Field, a free iPhone app of daily tracks in your own voice.
Why I built it
I used to record affirmations as voice memos and write them with ChatGPT. Listening back meant too many taps on my phone every morning. I wanted one place where I could record a track in my own voice and play it every day. Nothing did that, so I made it.
The stack, piece by piece
Every tool I pay for or depend on, grouped by what it does. Prices are what I actually pay as of October 30, 2026.
Build Where the app gets written.
- CodexAI coding agent. Wrote most of the Swift.$100 / mo
- ClaudeDesign, copy, debugging, second opinions.$100 / mo
- Swift + SwiftUINative iOS. No cross-platform layer.Free
- XcodeBuilds, simulator, TestFlight uploads.Free
- GitHubCode lives here. Every change has a history.$4 / mo
Run What keeps the app working after you close the laptop.
- RenderBackend API and Postgres database.about $40 / mo
- StoreKitIn-app purchases. Apple handles the money.15% of sales
- Apple Developer ProgramRequired to ship to the App Store.$99 / yr
Ship The public face.
- VercelHosts beinthefield.com, built with Next.js.$20 / mo
- GoDaddyDomains.about $20 / yr each
- Google WorkspaceEmail on my own domain. Customer calls get booked here.about $26 / mo
Sell Finding the people it's for.
- TikTok + InstagramWhere I talk about it. Short videos, not ads.Free
- Customer callsThe actual research. A calendar link and a conversation.Time
- TheannaAccelerator for growth experiments. Paid cohort.One-time
Business The paperwork.
- Ossur Labs LLCOne company. Be in the Field lives under it.$800 / yr in California
- An engineerPaid refactor once the app was real.Project fee
What it costs to run each month
| What | Per month | Only for the app? | Note |
|---|---|---|---|
| Codex (OpenAI Pro) | $100approx. | Shared with Ossur Labs | Writes most of the code |
| Claude (Anthropic Max) | $100 | Shared with Ossur Labs | Design, iteration, this site |
| Render | $40 | Yes | API servers + Postgres; was $14 before launch |
| Google Workspace | $26approx. | Shared with Ossur Labs | Business Plus, one seat |
| Vercel Pro | $20 | Shared with Ossur Labs | Marketing sites |
| Apple Developer Program | $8.25 | Yes | $99 a year |
| GitHub Pro | $4 | Shared with Ossur Labs | Private repos |
| Domains | $3approx. | Shared with Ossur Labs | A few on GoDaddy |
| Total | about $301 | $48 is the app alone | Before Theanna, the engineer, and the LLC’s $800 a year |
Most of that is two AI subscriptions. Take those out and the app itself costs less than a gym membership to keep alive. “Approx.” means I rounded from memory rather than a receipt.
Product
I started in Codex and built it as a native iOS app in Swift. No no-code builder, no cross-platform framework: I wanted it to feel like an iPhone app and I wanted to learn how one is made. The backend is a small API and a Postgres database on Render. Payments go through StoreKit, which meant an Apple Developer account, App Store review, and learning what Apple will and won’t let you do.
Claude did the design work with me and most of the iteration: screens, copy, the marketing site, the parts of the code I wanted a second opinion on. I used AI to help me launch, not just to build. Writing the App Store listing, debugging a rejected build, drafting the privacy policy: all of it went through the same two tools.
When the app was out and people were using it, I brought in an engineer. He looked at everything, kept what worked, and refactored the foundation so it would hold up as I keep adding to it. He did not rebuild from scratch. That’s the model I’m keeping: I build with AI, and he reviews when something matters.
Marketing
Going from zero to one is looking for signal. I don’t think there’s one right way to find it. Mine is a TikTok, an Instagram, and customer calls booked through a calendar link on my own domain. I put things out, talk to the people who respond, and see whether anyone will pay.
After launch I joined Theanna, an accelerator that started out helping women build businesses and now focuses on growing them. I use it to run growth experiments with actual numbers instead of guessing what to try next.
Legal
One LLC, Ossur Labs, and Be in the Field sits under it. California charges every LLC $800 a year, so one entity is the frugal choice until a product is big enough to need its own. The LLC owns the Apple Developer account, the domains, and the code.
Finance
Bootstrapped, on purpose. Low fixed costs, experts when the work needs them, and no spending ahead of signal. The one thing I haven’t done is write a business model, which you probably should do while you’re building, not after. That’s the next piece of this page.
When I brought in a human
- An engineer, after the app worked and people were using it. He refactored the foundation so it was safe to keep building on, and he didn't rebuild from scratch. I'll keep building with AI and have him review.
- An accelerator, once I had something to grow. Theanna runs structured experiments against real numbers, which is what I needed after launch, not before.
- Customers, from the start. Every call changed something. This is the only line here that costs nothing and the one I'd never skip.
What I haven’t done yet
- A business model. I know what it costs to run and what people pay; I haven't written down what has to be true for it to work. That's next.
- Android. Native iOS was the fastest path for me, and it's where my first users were.
- A team. It's me, AI, and experts when the work needs them.
Building your own thing and want to talk it through?
Start with a 30-minute call. Then, if it makes sense, a working session to get your idea clear and a plan you can act on. No equity, no pitch.
Questions people ask
Can a non-technical founder build an iOS app with AI in 2026?
Yes. I wrote almost none of the Swift by hand; Codex did, and I directed it. You still have to learn how an iOS app is put together, how to read what the AI gives you, and when to stop and ask someone who knows. The first 80% is fast. The last 20% is where the learning is.
What does it cost to run an app as a solo founder?
My whole stack runs about $301 a month, and most of that is two AI subscriptions I'd pay for anyway. The costs that belong only to Be in the Field, the backend and the Apple developer account, come to about $48 a month.
Do you need an engineer to launch an app?
Not to launch. I launched without one. You need one when real people depend on the app and you want to keep building on it for years. At that point a refactor by someone who has done it before is cheap compared to a rewrite later.
Why Render for the backend?
It was the simplest way to get a server and a Postgres database running without learning cloud infrastructure. It cost $14 a month before launch and about $40 after. When that becomes the wrong answer I'll know, because the bill or the latency will tell me.
Why one LLC instead of one per product?
California charges every LLC $800 a year whether it makes money or not. Ossur Labs is the company; Be in the Field is a product under it. If one product needs its own entity later, that's a good problem.