Should You Vibe Code Your Own Digital Signage Software

Open your laptop on a Saturday, point Cursor or Claude at an empty folder, type "build me a digital signage app that loops images on a TV," and by Sunday afternoon you have a working web page. It plays. It even looks okay on the Fire TV stick you borrowed from the back office. The urge to cancel the signage quote in your inbox is real.

That urge is the whole 2026 story. "Vibe coding" was named Collins Dictionary's Word of the Year for 2025, and operators who have never shipped a line of code are suddenly asking whether they should write their own menu-board software instead of paying for one. The honest answer depends on what breaks between day one and day thirty.

So what actually falls over between the demo and the lunch rush?

What Vibe Coding Actually Is in 2026

Andrej Karpathy coined the term earlier that year. His original framing emphasized full delegation to the AI, describing outcomes rather than reading or editing the implementation, and he scoped it explicitly to personal projects rather than production systems with real users. That scope matters, and most of the current discourse ignores it.

Developers themselves are all-in on the tooling. Stack Overflow's data shows AI tool adoption among developers climbed to 79% by 2025, with nearly half reporting daily use, and AI agent usage nearly doubled from 31% to 59% between the 2025 survey and an April 2026 pulse. But trust has not caught up with adoption. The 2024 survey already flagged a widening gap: 76% using or planning to use AI tools, only 43% trusting the output. Professionals use these tools and still read every line. That is not what a weekend vibe-coded signage project looks like.

The market pressure is real too. Gartner predicts that by 2026, 80% of low-code platform users will come from outside IT departments, up from 60% in 2021. If you own a dealership or run three gyms, you are exactly the person this prediction is about.

Day One Looks Great

Here is what you can genuinely build in a weekend with an AI assistant and no prior experience. A single HTML page that cycles through a folder of JPEGs on a timer. A simple JSON schedule that swaps the folder at 11:00 and again at 16:00. A Chrome window on a cheap Android TV stick, launched at boot by a sideloaded shortcut. Maybe a text file your manager can edit from a shared drive.

It plays. The images are sharp. The schedule flips. You feel clever, and you should, because five years ago that weekend would have taken a month.

Then Monday happens.

Day Thirty Is Where It Gets Expensive

The failures below are not edge cases. They are the ordinary mechanics of running screens in a real business, and they are what commercial signage software has spent fifteen years learning to handle. Vibe coding does not know what it does not know.

Offline Cache and the Blank Screen

Your Wi-Fi reboots at 10:52 on a Tuesday. The router is in the back. By the time anyone notices, the lunch board has been a blank black rectangle for twenty minutes and the line is out the door. Industry guides are blunt about this: a screen with no stored copy goes blank within seconds of a connection drop. A real signage player writes the playlist to local storage, keeps playing from that copy for as long as the line stays down, and does a delta sync when the network returns. Your vibe-coded page fetches images from a URL on every reload.

Scheduler Drift

The 11:00 switch from breakfast to lunch is the one job the system has to do right. The usual AI-generated version uses setTimeout or a browser interval, which quietly drifts after a few days, pauses when the tab is backgrounded, and resets on every reboot. A proper signage scheduler reads the device's real-time clock, survives power loss, and resumes the correct daypart on boot. If dayparts are central to your week, the mechanics are worth reading end to end in a dayparting best practices guide before you build.

Screen Fleet Recovery

One screen is a hobby. Four screens across two locations is a job. When the dispensary's second display freezes on a black frame at 7pm on a Friday, somebody has to drive there, find the HDMI stick behind the TV, unplug it, plug it back in, and hope. Commercial players ship with watchdog restarts, remote reboot, and health alerts for exactly this reason. Writing that yourself means a background service that monitors the renderer, a cloud endpoint that each screen checks into, and a dashboard that tells you which one is dark. That is not a weekend.

Android TV Packaging and Updates

Shipping a web page is easy. Shipping an Android TV app that auto-starts on boot, survives OS updates, and can be remotely updated across a fleet is a different project. Most serious deployments rely on an MDM layer: Android TV management platforms like Scalefusion, Hexnode, Radix and Esper exist specifically to handle silent app install, kiosk lock, and remote push across fleets, usually at $2 to $4 per device per month on top of whatever the signage software costs. Build it yourself and you own the enrollment flow, the signing keys, the update channel, and every 3am reboot.

Content Moderation and the Expired Promo

The quietest failure is the one nobody notices for three weeks: a 20% off coupon that ended on the 15th still playing on the 29th. Commercial systems handle this with content expiry dates, approval queues, and a scheduler that stops showing an asset after a cutoff. Your vibe-coded folder-of-JPEGs system does not know what October 15 means. The fix is a database, a review workflow, and someone whose job it is to check. All of which, again, is not a weekend.

The Security Side Nobody Talks About

This is where the research gets uncomfortable. A study by Escape.tech scanned 5,600 applications built with vibe coding tools and found over 2,000 vulnerabilities, 400-plus exposed secrets such as API keys and credentials, and 175 instances of exposed PII in production applications. The pattern behind that is structural, not accidental: vibe coding's speed-first, conversational workflow often bypasses pre-production security tools entirely, pushing unaudited code directly into production.

For a signage app showing a lunch menu, you might shrug. But the moment your screens pull in employee photos, patient wait times, loyalty data, or a POS feed, you have crossed into territory where an exposed admin endpoint on a public URL is a real problem. It is the same reason the questions to ask a vendor list is longer than most people expect.

When DIY Actually Makes Sense

None of this is an argument against building things. It is an argument for being honest about what you are signing up for. Vibe coding your own signage is a reasonable call when:

• You have one or two screens in a single location, and a blank screen for an hour is annoying but not expensive.
• Your content changes rarely, so the schedule is basically static.
• Nobody but you touches the system, so there is no approval or moderation workflow.
• You enjoy the maintenance, because the maintenance is the project.

It stops making sense the moment any of those flips. Multi-site, drive-thru, menu changes by daypart, more than one person uploading content, a boss who notices when a screen is dark. For a comparison of what the alternatives look like on the hardware side, the cloud signage versus media player box post walks through the trade-off in operator terms.

The Honest Build vs Buy Call

Think about what you are actually paying for when you buy signage software. It is not the loop of images. It is the fifteen years of edge cases somebody else already hit: the browser that stopped autoplay in 2019, the Chromecast firmware update that broke sideloaded apps in 2023, the daylight saving bug that turned a 7am open into an 8am open. Google's own framing of vibe coding is useful here, because even the companies selling the tools describe them as collaborators on prototypes rather than replacements for production engineering.

The build-vs-buy math is simpler than it looks. A pre-launch platform like AI Digital Signage charges a few dollars per screen per month and inherits all those scars. A weekend vibe-coded project charges you zero and bills you back in truck rolls, blank screens at lunch, and the Saturday you spend chasing a scheduler bug instead of running your business. If the premise of using AI to describe a screen in words and get a finished design back appeals to you, that is actually the thing commercial AI signage software is for. You get the vibe-coding feel without inheriting the ownership of the player, the fleet, and the 3am reboot.

Build the toy. Learn what the pieces are. Then decide. The screen in the window is a tool for your business, not a software project, and the test is not whether it runs on Sunday night. It is whether it is still running right the Tuesday after you stop paying attention to it.