Software we would have to run ourselves.
A web app MVP from A$19,500, a mobile one from A$29,000, an AI agent from A$7,500 and a chatbot from A$3,500 — ex GST, model costs passed through at cost. The proof is not a portfolio: it is two products of our own, in production, doing work every night.
The process, in five steps.
What you actually get.
A release you can charge for
Not a prototype. The smallest thing a real user can do the real job with, in production, with the admin you need to run it.
Cost ceilings on anything metered
Every AI feature we build has a budget in code, a refusal when it is reached, and an alert before it. We run that on our own free tools: each call is priced from the token usage the API reports.
Instrumentation from day one
First-party, no third-party analytics, no cookie you have to ask consent for. You get numbers you can act on and a privacy page you can stand behind.
Weekly, in the open
A deploy you can use every week, not a reveal at the end. It is also how scope gets cut honestly: you can see what you are not using.
The accounts are yours
Your cloud, your database, your API keys, from the first commit. We work in your accounts, not ours.
Handover or operation
We will run it for you with a named response time, or hand it over documented. What we will not do is make leaving expensive.
What it costs, from.
| What | From | Covers |
|---|---|---|
| Web app MVP | A$19,500 / US$12,900 | one product, one audience, shipped |
| Mobile app MVP | A$29,000 / US$19,000 | one platform to start |
| AI agent or automation | A$7,500 / US$4,900 | one workflow, end to end |
| AI chatbot | A$3,500 / US$2,300 | running costs at cost |
Each covers one platform and one audience; the second mobile platform is 60 hours because it is a second app. Hourly work is A$140 / US$95, and model usage is passed through at cost with a ceiling you set. AUD is ex GST for Australian clients; USD everywhere else. Every one of these is a fixed price against a written scope before the work starts — and the cost guide shows the arithmetic, including the hours behind every addition.
Two products of ours, in production.
Bilby is an AI customer: every night it visits a store from six regions in a real browser, on desktop and a phone, from the homepage through to the payment step, and reports what broke with screenshots. It is a queue, a fleet of browser workers, a judging step on a language model with a per-call cost ceiling, an alerting path and a billing system — which is to say it is the kind of thing this page is selling, and it is running right now.
Navaal: AI SEO, AEO & GEO is a published Shopify app: OAuth, webhooks, GDPR data requests, Shopify billing and an App Store review passed. If you are building on a platform with a review process, that experience is most of what you are buying.
Straight answers.
What counts as an MVP in your prices?
The smallest version a real user can do the real job with: one platform, one audience, the one workflow that matters, and the admin to run it. Not a demo, not a prototype. Everything else is a second release, priced when the first has told you something.
Who owns the code and the accounts?
You, from the first commit, and we work in your accounts rather than ours. That is not generosity: it is the only arrangement where you can leave, and a studio that makes leaving expensive is a studio with a reason to.
How do you keep AI running costs under control?
A budget in code, enforced before the call, priced from the usage the API actually reports — and a refusal that says so when the ceiling is reached, rather than a silently worse answer. Our own free generator runs on a US$2 monthly cap with five runs per visitor per day, and the page says what happens when either is hit.
Can you build on top of an existing system?
Usually, and the first question is what it can tell us. If it has an API or a database we can read, yes. If the answer is “a screen somebody copies numbers off”, that is still yes but it is a different job, and we will say which before quoting.
Do you do mobile apps for iOS and Android?
Yes, and we will ask whether you need both in version one. The second platform is 60 hours — A$8,400 — because it is a second app with its own store review. Shipping one platform to the audience you actually have is usually the better first move.
What if the project should not be built?
We will say so in the first conversation, which is free. We would rather lose a job than bill for one that cannot pay for itself — partly because the reference matters more to us than the invoice.
Prices are published. A quote still needs a conversation.
The cost guide prices your own build from the same hours and the same rate we quote with, and the form at the bottom of it carries your estimate into the message — so the reply starts from what you chose.