What to vibe code first

What to vibe code first is not Notion. The catalog lists dozens of subscriptions you could replace. The ones people click first are the ones they resent the most.

Notion. Slack. Figma. The workspace that ate the notes. The chat the company lives in. The file three people have open at once.

I get why. Those are the charges you notice.

They make a poor first project.

A useful Notion replacement, the kind that covers what you actually do in it, is a weekend if you scope hard. The page says a basic version takes about 10 hours. Matching everything Notion really does is closer to a month. Slack and Figma sit in the same bucket: side quests, days of work, and a long list of things you should not build in v1.

You will still open the paid app on Monday, because the notes are still in there.

What to vibe code first is a tonight project

The first replacement should be a tool you can put on a URL before you get bored.

Look at the effort labels on this site. Tonight. Weekend. Side quest.

Tonight means a basic version in a couple of hours or less. You get something you can use. You do not get a marketplace, a billing system, or every integration the paid product spent a decade adding.

Each product page already splits the work into three rows.

The first row is the basic version. A list of links. Compress and download. A short URL. A form and a table. A status board.

The last row is the honest rebuild. Multi-tenant hosting. A theme marketplace. Every network API. A visual builder. That row is there so you skip it.

If you tell a coding agent to “clone Typeform”, it will aim at the last row. Use the prompt on the page. It is written for the first one.

Five tiny subscriptions I would cancel first

These are all marked tonight on the site. The prices are the public paid plans already in the catalog. The times are the basic-tier estimates, not the fantasy rebuild.

Linktree, $9–$24 a month

About 30 minutes for a links list, an avatar, and one theme.

That is the whole product most people use. The same v1 also covers Beacons and Bio Link. Yearly that is $108–$288 for a page of URLs.

If you have been paying rent on a list of links, build this one first. The 30-minute walkthrough covers that v1. The 2-hour version adds drag-to-reorder, drafts, and daily click counts. Stop there. The week-long version is a hosted platform with billing and a theme marketplace. You do not need that.

TinyPNG Pro, $25–$39 a month

About 60 minutes to drop images, compress them, and download them.

Do the work in the browser so the originals never leave your machine. The page is explicit about that. No accounts. No CDN. No “image platform”.

$300–$468 a year is a lot for a quality slider. The replacement is a drop zone and a download button. The TinyPNG alternative in the browser is the 60-minute path. The 3-hour version adds a side-by-side preview and a zip export, which is the moment it feels finished.

Bitly, $35–$300 a month

About 90 minutes.

A short URL and a click count. The catalog calls it tiny by definition, and the price range tells you what happened after that. The low end is already $420 a year. The high end is $3,600 a year for redirects.

Build the redirect and count the clicks. Stop before someone asks you for UTM workspaces and QR campaigns as a service.

Typeform, $29–$99 a month

About 2 hours for a one-question-at-a-time form and a response table.

Most people need seven questions and a thank-you screen. Text, email, a select, a long answer. An email when someone submits. A CSV when you want the rows on your machine.

That v1 also replaces Jotform and Youform, because they all do the same job. The 6-hour version adds logic jumps, theming, and a webhook. The visual builder and payment fields can wait until you personally miss them, which for most of us is never.

$348–$1,188 a year is a lot of money for a questionnaire.

Statuspage, $29–$399 a month

About 3 hours.

A public page that says whether your stuff is up. The catalog calls it a perfect tiny app, and the $399 top end is the giveaway. You do not need an incident-command platform to tell people the API is slow.

A status board is a list of services, a current state, and a short history. Host it next to the product it describes. Update it when something breaks. That is the job.

If you care about the money more than the evening

DocSend is still only about 4 hours, and the public plans are $65–$180 a month. The site marks it a huge savings candidate for a reason. A link, a password, and a view count. That is the useful product.

Do it second, not first, if you have never shipped one of these. The first win should be small enough that you finish it. DocSend is the one you reach for when you already believe the loop works and you want the charge that actually hurts.

Carrd sits at the other end. $2–$4 a month, about 2 hours, one page. Cheap enough that the savings will not change your year. Still a clean first ship if a personal site is what you have been renting.

How to use a product page

Open the one you picked. Read the audience line and the non-goals before you copy anything.

The audience is who the replacement is for. One creator. A small business collecting leads. You, not “teams everywhere”. If that does not sound like you, pick a different card.

The non-goals are the part most people skip. They are the most useful sentences on the page. They tell the agent what not to invent.

Then pick the stack you already know. Every prompt ships in four flavors: AHA (Astro, HTMX, Alpine.js), Next.js, Laravel, and Rails. The AHA versions target Cloudflare D1, R2, and Workers when the product needs them. This site is an Astro app on Cloudflare, so that path is the one I default to. Use Rails if that is what you write every day. The scope matters more than the framework.

Copy the prompt. Run it. Deploy the thing. Use it for a week as the real tool, not as a demo.

Then cancel.

If you miss a feature, add that feature. Do not add it because the paid product has a button for it.

The product pages also have mini-prompts under “Build just one piece”. Those are for the weird sub-problem inside a bigger app, an evening with no promise that you will replace the whole subscription. Fine as a side path. For a first cancel, take the basic v1.

When you should keep paying

Keep the tool that is someone else’s source of truth.

Slack, if the company already lives there. Figma, if three people are in the same file. Notion, if the rest of the team will not follow you to your rebuild.

The homepage says this without dressing it up. Keep the hard stuff. Build the tiny stuff. Cancel with gusto.

A personal link page is tiny. A compressed-image drop zone is tiny. A status board is tiny. A form with a response table is tiny.

Per-person pricing changes the math, and not in your favor. Notion is $12–$24 per person per month. Slack is $15–$18. Figma is $16–$55. The audit on the homepage lets you type a team size, because that is how those vendors charge. A focused rebuild for one person can still make sense. A rebuild you have to sell to twelve coworkers is a different project.

Hosting should not scare you off the tiny ones. The product pages already put it this way: a focused replacement costs your build time plus close-to-zero hosting. That assumes you are not rebuilding the marketplace.

What to do tonight

Go to the audit. Filter by Tonight. Pick the one you already pay for.

If none of those are on your bank statement, look at QR Tiger or Shots.so. Same shape. Small surface. A done state you can see.

Notion will still be there when you want a weekend project. Ship the link page first.