NavigateMENU / 04
More
Elsewhere
Offer · Paper Crane Starterv1 · 2026
Three months to launch.
Yours to extend.
Two Shape Up cycles, end to end. The same Postgres database that runs Mobbin, the same Vercel platform that runs Sonos. Deployed as your codebase, on your accounts. Built faster because the substrate is solved before we start.
Want the speed multiplier? Read how AI-driven development fits underneath.
How the three months unfold
Two Shape Up cycles, end to end.
Cycle 1 · Weeks 1 to 6
Setup, integrations, strategy
Discovery wrap, design system tuned, accounts wired (Supabase, Vercel, GitHub), first integrations in, and the first feature lands in production.
Cycle 2 · Weeks 7 to 12
Build out and ship
Remaining feature surface, polish, QA, and the launch. End of cycle 2 is a working MVP in users' hands, not a staged demo.
What you get
A solved substrate. A launched product in two cycles.
We deploy a per-client Next.js + Supabase substrate with the patterns every modern application needs already in place. The two cycles go to the things that make your product different, not to rebuilding authentication for the hundredth time.
Authentication and access
Role-based access, row-level security, invite-only signup, audit-grade impersonation for admins. The permission model auditors actually want to see.
Customer portals
Magic-link client portals that don't require a full account. Per-company opt-in, session revoke, rate-limited so a stolen email can't farm logins.
Communications
Transactional email with delivery and open tracking logged to your database. Scheduled email queues. Slack treated as a first-class peer to email, not a bolt-on.
Operations tooling
Cron endpoints for nightly maintenance. Background jobs with status records. Audit trails, undo patterns, an internal bug-report channel. The unglamorous work that keeps software trusted.
Integrations
Google Drive, Calendar, and Contacts. Xero. Slack. GitHub. Documenso for e-signatures. Monday.com. Wired in once, reusable on every project.
Velocity
Forty-plus feature modules to clone from. A form framework that cuts boilerplate by roughly 70 percent. A typed database layer so the compiler catches schema mistakes before users do.
Why this stack
Real code. Real ownership. Real ceiling.
Next.js is what Mobbin, Sonos, and Notion ship on. Supabase is Postgres in a clean wrapper. Together they replace the four-or-five vendor stack most agencies still pitch, and they do it without locking you into a proprietary format.
When something needs fixing, you read your codebase. When you outgrow us, the repo, the database, and the hosting accounts are yours to take anywhere. That's the test we apply. If Paper Crane disappeared tomorrow, would the buyer still own their software? Yes.
WordPress is genuinely good at what it was built for, which is content. It gets shaky as the foundation for an operational stack: a WordPress site plus a Zapier flow plus a form builder plus a CRM, held together by middleware. The chain is only as steady as the slowest vendor in it.
Clean stackone substrate, one direction
Stitched stackfive vendors, six handshakes
PC Starter Next.js + Supabase | Custom dev shop Quote-based | Low-code Bubble, Webflow, Retool | Frankenstein WordPress + plugins + Zapier | |
|---|---|---|---|---|
| Build time | Three months. Two Shape Up cycles | Six to twelve months | A weekend, then a wall | Quick to stitch, slow to consolidate |
| Run cost | Under $100/mo to start. $200 to $600/mo at scale | $1k to $15k/mo with DevOps quietly included | $30 to $300/mo, until the scale tier | Plugin subscriptions, stacked |
| Code ownership | Full. Repo transferred to you | Varies by contract. Often not | Proprietary. No export path | Core is yours. Plugins rent forever |
| Scale ceiling | Ten million users on the same stack | Depends entirely on what you bought | Hard caps. 10k CMS items, 50k characters of custom code | Steady until a vendor in the chain changes the rules |
| Time to fix something | Hours. You have the code | Days to weeks. Back in the agency queue | Impossible if the platform doesn't support it | Depends on the plugin author's release cadence |
What it costs to run
The numbers, not the marketing.
Modern infrastructure flipped what a working app costs to keep alive. The first three months get you a useful MVP that covers a focused subset of use cases, not a full platform. Here's the breakdown, and what it replaces.
$50 to $100
Monthly run cost at the start
Supabase Pro ($25), Vercel Pro ($20), a domain, and headroom for traffic. That's a working SaaS bill, not a trial tier.
$200 to $600
Monthly run cost at mid-scale
Past 100k monthly active users with real traffic. Same stack. No re-platform required.
3 months
Time to a launched product
Two Shape Up cycles, end to end. The kit takes the substrate off the table on day one, so the cycles go to your product, not to rebuilding auth, email, and customer portals. Traditional shops quote six to twelve months for the same scope.
1 to 2
Engineers required
Full-stack on the modern stack. Traditional shops quietly fold a $125k to $180k DevOps salary into the line items.
Build cost for the two cycles
- $5,000 to $8,000 per month
- $15,000 to $24,000 total over two cycles
Final number depends on scope and pace. We give a firm range after the discovery call. Same range we quote on the where-to-start track.
After the MVP
How we work past launch.
We tend to stay involved on an ongoing basis. Retainers run $4,000 to $9,000 per month depending on how fast the team wants to ship and how the app is growing. Some clients drop to dependency updates only; others scale up to add new flows every cycle.
30-day cancellation terms. Ready to step off the retainer? Let us know and we wind work down with near immediacy.
Run costs are real numbers, not headline numbers. They cover Supabase, Vercel, and a domain. They don't include the work to build your product on top, which we quote per engagement.
Not for
Honest about what we won't do.
This kit is wrong for some projects. We'd rather you find that out here than five weeks in.
Native mobile apps
This starter ships web. Responsive web that feels native on a phone, yes. A Swift or Kotlin app in the store, no, not from this kit. We do build native mobile, just through a different engagement. Happy to walk you through how that one looks.
Government and defense-grade compliance
Supabase carries SOC 2 Type 2 and HIPAA with a BAA. It does not yet carry FedRAMP, ITAR, or ISO 27001. If you're selling into public sector or regulated defense, the substrate is wrong.
Sustained heavy compute
Video transcoding pipelines, large-scale ML training, real-time multiplayer at 100k concurrent users. Serverless billing punishes long-running CPU. We have alternative approaches that may fit, and we're happy to walk through them.
The speed multiplier
Built faster, with AI in the loop.
The kit gives the codebase a known shape. AI-driven development takes that known shape and ships against it faster than a team starting from scratch. Same patterns, same modules, same tests, all of them legible to the model. The gap between a typing-tutor AI workflow and one that actually lands working features in a sprint is whether the substrate is solved.
Where to next
Tell us what you're trying to build.
If the page has done its job, you already know whether this fits. The next step is short. Send us the shape of the problem, the systems you're already using, and what working looks like for the team. We'll tell you whether the kit is the right starting point, or honestly point you at something else.
Start the conversation