FAQ

Answers to the questions that come up most on eval calls.

Questions we hear on almost every discovery call. If yours isn't here, contact us and we'll add it.

Pricing

Do I need my own Claude API key?

Free tier: no. You use Claude via your existing Claude.ai account (in Claude Code / Cursor / Codex) — the AI runs read-only against your fork.
Professional tier: no. We provide the AI agent's access as part of the subscription. You still use Claude Code / Cursor / Codex — the difference is the AI has write access to your fork under your team's supervision.
Enterprise tier: yes, BYO Claude API key. This gives you full quota control and keeps AI usage on your account for billing / compliance reasons.

What's included in the $5K setup fee?

Business discovery session (60 min): what you're building, why, who for
Configuration session (90 min): fork core, wire deploy, set brand
First template install session (90 min): live install + first deploy
Two follow-up sessions in the first two weeks to unblock issues
Slack access from day one

Fee is waived for annual commitments.

Can I cancel?

Yes. Monthly commitments cancel with 30 days' notice; annual commitments cancel at the end of the term. You keep everything — code, deployments, data. See Ownership, hosting & security.

Why $500+/month for Professional?

Because you're getting a small engineering team's output at a fraction of one engineer's salary. A senior full-stack engineer costs $180k–$250k/year loaded ($15k–$21k/month). Professional tier gives you roughly 60–80% of one engineer's throughput via the AI + our weekly session, for a small fraction of the cost.

If you're building something where 60–80% of one engineer isn't enough, you either need Enterprise (dedicated support) or you need to be honest that you need a real engineering team.

Product

What templates are available today?

Five: SaaS Starter, Dashboard, AI Chat, Landing, Customer Portal. More shipping weekly. See /templates for live demos, or Templates for the docs.

Which AI agent do I use?

Claude, via one of: Claude Code (recommended for most), Cursor, or OpenAI Codex. All three work — you pick your preferred IDE, we give you the setup prompts + MCP configuration.

Can I use my own domain?

Yes. You bring the domain, we help you point DNS to your Cloudflare Pages projects. Custom subdomains, apex domains, wildcard — all supported.

Do you support anything other than Nuxt / Cloudflare?

Today: Nuxt 4 + Cloudflare Pages is the paved path. Enterprise tier gets AWS / GCP support and can bring their own hosting stack. We do not support other frontend frameworks (React, SvelteKit, etc.) — the whole platform is Nuxt.

How is this different from Lovable / Bubble / Cursor?

Vs. Lovable / Bolt / v0: those give you a .zip and disappear. We give you a live monorepo, ongoing agent access, and a human team. Their output is toy-scale; ours is production-scale.
Vs. Bubble / Webflow / no-code: those don't give you real code. We do. If you outgrow them, you rebuild from scratch. If you outgrow us, you take the code and keep going.
Vs. Cursor / Claude Code alone: those are the tools we run on. Alone, they need a technical operator to use effectively and a codebase to work in. We give you both — a codebase already structured for AI development, and the humans who know how to operate it.

Team & process

How often do we meet?

Professional: weekly, 60 minutes
Enterprise: your choice (daily, bi-weekly, monthly)
Ad-hoc: Slack async during business hours

Who's on the App-Agent team?

Small team of founders + operators. Same people you talk to on your discovery call are the same people who show up to your weekly sessions. No handoffs to a support tier.

What if the AI ships bad code?

You review and merge. Nothing hits main until you approve it. During your first few sessions we walk through how to review PRs from an AI agent — spotting the common failure modes, running tests, catching regressions. After a few weeks most customers are comfortable reviewing agent output solo.

What if I want to hire a real engineer later?

Great. The codebase is standard Nuxt 4 — any Nuxt-familiar engineer can read and edit it. Some of our customers use us to bootstrap the first 3–6 months, then hire in-house and keep the AI as a supplementary tool. Others use us long-term as their engineering function.

Next steps

Getting started

What happens after you book a call. Session cadence, what to prepare, first-week expectations.