The AI tool graveyard: how to spot a dying tool before you build on it

For every AI tool that becomes a household name, dozens quietly die. They run out of funding, get acquired and shut down, have their one feature absorbed by a model provider, or slowly rot — no updates, rising prices, support that stops answering. The tool directories will never tell you which ones are dying, because their business is promoting tools, not burying them. A directory that only celebrates winners isn't a directory. It's a sales brochure.

That leaves you to read the signs yourself — before you build a workflow, a product, or a business on something that won't be here next year. This is the field guide the directories won't write: why AI tools die, the warning signs that show up early, and how to protect yourself from betting on a ghost.

How AI tools actually die

Death comes in a handful of recognizable shapes. Knowing them helps you see it coming.

• The feature gets absorbed. A tool does one useful thing on top of a model — and then the model provider ships that exact thing natively. Overnight, the tool's reason to exist evaporates. This is the most common AI-specific death.
• The funding runs out. A venture-backed tool priced below cost to grow, never found a business model, and the runway ended. The app keeps working for a while, then the servers go quiet.
• Acquired and killed. A bigger company buys the tool for its team or technology and shuts the product down. Your data and workflow go with it, usually on a short notice period.
• Enshittification. The tool doesn't die so much as decay: the free tier vanishes, prices climb, the best features move behind an enterprise wall, and the thing you loved becomes something you tolerate, then leave.
• Quiet abandonment. No announcement, just entropy. Updates stop, the docs go stale, bugs pile up, support stops replying. It's technically alive and functionally dead.

The warning signs (read these before you commit)

Dying tools signal it early if you know where to look. Run down this list before you build anything important on a tool — and again every few months for the ones you already depend on.

• Stalled updates. Check the changelog, release notes, or GitHub activity. A tool that hasn't shipped anything in months is coasting, and coasting precedes stopping.
• Pricing moving the wrong way. Free tier removed, prices hiked, features shifted to enterprise-only. These are the moves of a company trying to survive on its existing users rather than grow — a late-stage signal.
• Founder or team exodus. Key people leaving, especially quietly, is often the earliest real signal. Watch who's still shipping and speaking for the product.
• An acquisition with no roadmap. "We're joining [BigCo]" with no clear commitment to keep the product running usually means a wind-down is coming.
• Community mood souring. Rising complaints, unanswered support threads, people openly asking for alternatives. The community feels a death before the company admits it.
• Docs and support rot. Outdated documentation, broken links, a support inbox that goes silent. Neglect of the unglamorous parts is a reliable tell.

The community always feels a tool dying before the company announces it. Rising complaints and "anyone know an alternative?" threads are the obituary written in advance.

How to protect yourself

You can't avoid every dying tool, but you can make their death survivable. The whole game is refusing to be locked in.

1. Prefer tools you can leave. Favor ones with data export, open formats, and standard interfaces. If leaving is a click, a shutdown is an inconvenience, not a catastrophe.
2. Keep your data yours. Export regularly. If a tool holds something you can't get out, you don't own it — the tool does, and so does whoever acquires it.
3. Favor tools with a real business model. A tool that charges a fair price and clearly makes money is far more likely to be here next year than one burning venture cash to stay free.
4. Abstract the critical dependencies. If a tool sits at the center of your product, wrap it so you can swap it. The model or service being replaceable is what lets you survive its death.
5. Re-check the signs on a schedule. Once a quarter, run the warning-sign list on the tools you depend on. Migrating early and calmly beats migrating in a panic when the shutdown email lands.

Why nobody else publishes this

It's worth saying plainly. The big AI directories run on affiliate deals and paid placement — some literally auction front-page rank to the highest bidder. Their revenue comes from the tools they list, so they structurally cannot tell you a tool is dying, overpriced, or a bad bet. The honest version — which tools to avoid, which are decaying, which just got worse — is exactly the content their business model forbids. That's the gap this exists to fill.

Common questions

How do I know if an AI tool is going to shut down?

Watch for stalled updates, pricing moving the wrong way (free tier removed, hikes, enterprise-only features), team departures, souring community mood, and rotting docs or support. No single sign is proof, but two or three together mean it's time to plan an exit — while it's calm, not when the shutdown email arrives.

Why do so many AI tools die?

The boom produced an enormous number of tools built on the same models, many with no real moat or business model. When a model provider absorbs the feature, or the funding runs out, or a bigger company acquires and shelves it, the tool goes. High birth rate, high death rate — it's the shape of a gold rush.

The takeaway: AI tools die constantly, and the people ranking them won't warn you. Read the signs yourself — stalled updates, wrong-way pricing, team exodus, souring community — and protect yourself by refusing lock-in and keeping your data portable. Build on things you can leave. More honest tooling guidance on the Kapyn Radar.