Most AI tool directories are dead, dying, or lying. I'd like MoreAI.tools to be none of those. Here's how.
What "dead, dying, lying" means
While researching the directories that already exist, I checked 34 of them. Of those:
- Two were literally offline. Domain expired, hosting lapsed, nothing to recover. AI Scout and GetAI.tools are the visible ones; there are more that I couldn't reach at all.
- At least eight had front-page tool listings that linked to dead vendor pages. The tool itself was acquired, shut down, or pivoted, and nobody on the directory side noticed.
- At least twelve had stated pricing that was off by more than 25% from the vendor's current page. Most were stale by 6 to 18 months. One was off by 2x in either direction.
- At least four were quietly running "top picks" sections sorted by affiliate payout, with no disclosure that the ranking was paid.
These aren't intentionally bad sites. They're the natural failure mode of "set it up once and let SEO juice carry it." The directory exists to harvest a few hundred dollars a month in affiliate revenue, and once it works, nobody updates it.
The result is: when you Google "best AI tools for [your job]" and click the top organic result, there's roughly a 1-in-3 chance it tells you to use a product that doesn't work the way it's described, costs twice what's listed, or doesn't exist anymore.
That's the bar. We need to be better than that. Here's the pledge.
The pledge, line by line
1. We've used every tool we list
Not demoed. Used, in real work, for real outputs. The Stack Score reflects that judgment.
There are exceptions for category coverage — sometimes we'll list a tool that's a major player we haven't personally used (DeepSeek would be one), but those carry a lower Stack Score and the founder POV says "We haven't used this in production. Here's what we know."
2. Pricing is verified weekly
Every tool has a last_pricing_change_at timestamp visible on the detail page as a "Pricing verified [date]" badge. Behind it is a Monday-morning pipeline that:
- Curls every tool's pricing page with a real browser user-agent.
- Falls back to a headless Chromium for the ~15 sites that block automated requests.
- Feeds the cleaned page text through a structured LLM extraction.
- Updates the database where the new tier data differs from what's stored.
- Stamps the date.
It's not bulletproof — vendors who change pricing mid-week will still have a 6-day stale window. But the worst-case staleness is one week, not 18 months.
3. Editorial scores are set BEFORE affiliate revenue
When we add a tool to the catalog, we set its Stack Score and write the founder POV before we ever check whether there's an affiliate program available, and before we apply to one.
We then apply to the affiliate programs that match. Sometimes we don't apply because the program is gross. Sometimes we apply and get rejected for traffic. Sometimes the tool has no affiliate program at all — Midjourney, ChatGPT, Claude, GitHub Copilot, and Figma are all in that category, and they all sit at 80+ Stack Scores. The score isn't downgraded for "we don't get paid for this."
If we ever launch a sponsored slot on the site, it'll be labeled "Sponsored" with the same UI as the existing badges on the tool detail page. The slot will be limited to tools we'd recommend anyway. We will never re-rank a non-sponsored tool down to put a sponsored one on top of it.
4. Bias is disclosed, not hidden
The editor (me) tilts toward:
- Claude over ChatGPT for long-form writing. (Both are 92.)
- North American workflows because that's my market.
- Tools with vocal founders over tools with corporate marketing.
These tilts affect scoring. The /how-we-score page lists them publicly. When they're relevant to a specific tool's review, the founder POV says so out loud.
5. We say what we don't list, and why
There's a /blog/what-we-dont-list-and-why post that names every tool we considered and skipped, with the reason. We update it as new tools surface.
This is the post most directory operators wish they could write and don't, because the entries make you specific enemies. We'll take that hit. The signal value is worth more than the goodwill.
6. Tools that go dark get unlisted
When a tool is acquired, shut down, or pivots so substantially that the listing no longer reflects what's offered, we flip verified_status to "unverified", which surfaces a warning on the detail page. After 90 days unverified, the tool gets removed from the main listing and moved to a /removed-tools archive page (with the date and reason logged).
This is also a side effect of the quarterly rich-profile re-test — every 90 days we re-verify the top of the catalog.
7. If we break the pledge, call us out
Reply to any newsletter. Email the editor. Use the contact form when we have one. If something's been re-scored to elevate a sponsor, or pricing is stale, or a deprecated tool is still listed, we'll fix it within a week or explain why we won't.
What this isn't
This isn't a pledge of objectivity. Editorial work isn't objective. The whole value of a curated directory is that someone's making opinionated calls. The trust isn't "we have no opinions" — it's "you can see the opinions clearly enough to decide whether they match yours."
If our biases don't match your context (you're outside North America, you prefer ChatGPT's ecosystem, you live in a different toolset entirely), the right move is to read our scores critically and discount accordingly. We'll never resent that.
The short version
We've used the tools. The pricing is current. The scores are set before the money. We say what we skip and why. We unlist dead tools. And if we break any of it, we'd rather hear about it than not.
That's the bar. Hold us to it.
— MoreAI.tools editor