L25 · Setup · 5 min
What it costs, and how to avoid surprises
In one sentence
Building your first apps requires one paid subscription—your coding agent—while GitHub, Vercel Hobby, Neon and Sentry start on free tiers, and any AI features inside your app use prepaid credit with daily spend caps and separate low-limit development keys so you never face a surprise bill.
Think of it like this
Think of a prepaid mobile plan with a daily limit instead of an open tab. You decide upfront how much credit to load, set a hard ceiling per day, and get a text alert long before the balance runs low.
How it works
Your only fixed monthly bill when starting out is your Definitioncoding agentAn AI that works in your project folder: it reads files, writes code and runs commands, with your permission.Open in the glossary → subscription (such as Claude Pro or higher). Everything else you need for a personal project starts free: GitHub stores your Definitionrepo (repository)Your project folder plus its full history. Git records each checkpoint so you can go back to any version.Open in the glossary → for free, [[vercel]] Hobby hosts personal, non-commercial sites for free (switch to Pro when you charge money or run a commercial project), and both Neon (DefinitiondatabaseWhere your app remembers things between visits, like accounts or saved recipes.Open in the glossary →) and Sentry (error alerts) offer generous free plans.
Each app gets only the pieces it needs: if your app doesn't store data between visits, you don't open a Neon account; if it doesn't have AI features, you don't pay for an AI provider. You can review the full breakdown at any time on the What it costs page.
When your app does call an AI DefinitionmodelThe AI system that produces answers, like the one behind your coding agent.Open in the glossary →, you avoid open-ended bills in four ways: (1) load prepaid credit on Tokentide (starting from $10, which never expires) and Vercel AI Gateway; (2) turn on a daily spend cap and email alerts in each dashboard; (3) create a separate low-cap [[api-key]] for .env.local on your computer so a local loop can't touch your Production budget; and (4) set max_completion_tokens and a Definitionrate limitA cap on how often something can be done, like sign-in attempts per minute, to stop abuse.Open in the glossary → on every AI endpoint.
What it looks like in your project
Your agent always asks your permission before installing paid services or spending money. When a project uses AI, your agent adds max_completion_tokens, rate limits, and server logs that record X-Tokentide-Charge-USD so you can see the exact cost of each request.
Copy for your agent
Review my project and list every external service it uses or plans to use, whether each one is on a free tier or paid, and what caps and rate limits we have in code to prevent unexpected spend.
Quick check · 1 question