Free cookie consent management tool by TermsFeed Generator Update cookies preferences

The AI tool graveyard: 8 I cancelled and why

The AI tool graveyard: 8 I cancelled and why
Photo by Rodion Kutsaiev / Unsplash

There's a version of this site that only covers tools I recommend. That version would be shorter and more comfortable to write.

It would also be less useful to you.

Knowing which tools are worth paying for is valuable. Knowing which ones aren't is arguably more valuable — because the AI tool market is full of well-marketed products that sound compelling until you actually use them. The cancellations tell you something the launches don't.

Here are eight AI subscriptions I've cancelled in the past eighteen months, what each one promised, and the specific reason it got dropped. I'll be more useful about this than most cancellation lists, which tend to say 'it just wasn't for me.' I'll tell you the actual problem.


Knowing which tools aren't worth paying for is more valuable than knowing which ones are. The AI tool graveyard is where the honest signal lives.

Jasper  —  £39/month

Was using it for: Long-form blog drafting and newsletter content

Why I cancelled: Jasper produces fluent, grammatically correct prose that sounds like no one in particular. The output is competent and generic in equal measure. I kept trying to use it as a starting point and kept rewriting everything from scratch anyway, which made the subscription pointless. When Claude became capable enough for drafting, Jasper had no remaining argument for its price.

Copy.ai  —  £29/month

Was using it for: Short-form copy — headlines, subject lines, ad copy

Why I cancelled: Copy.ai is fine at what it does, which is producing variations quickly. The problem is that it does the same thing Claude does, just less well, for more money. The 'workflows' feature that justifies the premium never became part of how I worked. Cancelled after three months when I realised I hadn't opened it in six weeks.

Writesonic  —  £19/month

Was using it for: SEO article drafts with built-in keyword optimisation

Why I cancelled: The SEO features are the pitch. The reality is that Writesonic produces stuffed, mechanical content that reads like it was written for a crawler rather than a person. I use Frase for SEO research and Claude for drafting — the combination does what Writesonic promises to do, but better. No reason to pay for the middle option.

Otter.ai  —  £17/month

Was using it for: Meeting transcription and summary

Why I cancelled: Accurate enough transcription, but I don't attend enough meetings to justify the cost. The summary feature was useful the handful of times I used it but never became essential. Cancelled after four months. If you're in back-to-back meetings all day this is probably worth it. I'm not.

Grammarly Premium  —  £12/month

Was using it for: Grammar checking and writing suggestions

Why I cancelled: Grammarly Premium is a perfectly good product that became irrelevant once I was using Claude for editing. The suggestions overlap substantially and Claude's edits are more contextually intelligent — it understands what you're trying to say, not just whether the sentence is grammatically correct. The free tier of Grammarly is still occasionally useful in email. The premium tier isn't.

Lumen5  —  £19/month

Was using it for: Turning blog posts into short social videos

Why I cancelled: I don't use social media for any of my publishing projects, so this was always a speculative subscription. The output was better than I expected. It didn't matter because I had no distribution channel to use it. Cancelled after six weeks. If you're actively producing social video content this might be worth a look — but test the free tier properly before paying.

Rytr  —  £9/month

Was using it for: Short content generation — emails, product descriptions, social posts

Why I cancelled: Rytr is cheap, which is its main selling point. At £9/month it costs almost nothing. It also produces almost nothing worth keeping. The outputs are template-following rather than thinking — you can see the structure of the prompt in the output. Upgraded to Claude and never looked back.

Surfer SEO  —  £49/month

Was using it for: On-page SEO optimisation and content scoring

Why I cancelled: This is the most expensive cancellation on this list and the one I thought hardest about. Surfer SEO is genuinely useful software — it scores your content against top-ranking pages and tells you what to add or cut. The problem is that at £49/month for a solo creator it's hard to justify against Frase at £14/month, which does the core job well enough. If you're running an agency or a large content operation, Surfer probably earns its keep. For a one-person publishing business, it doesn't.

What the cancellations have in common

Looking at this list, a few patterns emerge.

Duplication without differentiation

Most of these tools do something that another tool in my stack already does. Jasper and Copy.ai both do what Claude does, less well. Writesonic does what Frase plus Claude does, less well. Grammarly Premium does what Claude does, less well. When you have a strong core tool — in my case Claude for writing and Frase for SEO research — the category around it collapses. There's no room for a slightly worse version of the same thing at a positive price.

Features I used once

Several of these tools had features that impressed me during trial — Surfer's content scoring, Lumen5's video output, Otter's summaries — but that I only used a handful of times in practice. A feature that impresses you in a demo and that you use once a month is not worth paying for. The question isn't whether something is impressive. It's whether you'll use it on a Tuesday afternoon three months from now.

The trial problem

Most of these tools offer free trials. The problem with free trials is that you use them differently to how you'd use a paid subscription. During a trial you're evaluating. After a few weeks of paying, you're relying. Tools that perform well under evaluation often fail under reliance — either because the novelty wears off, or because the actual daily workflow doesn't match the use case you were testing.


A feature that impresses you in a demo and that you use once a month is not worth paying for.

How to avoid this — how to trial a tool properly before you pay — is the subject of Friday's post.

My current stack — what survived the graveyard: https://thepracticalai.digitalpress.blog/my-ai-stack-2026/

Why most AI tool reviews miss this entirely: https://thepracticalai.digitalpress.blog/why-most-ai-tool-reviews-are-useless-and-what-this-blog-does-instead/

— Ellis

About Ellis

Ellis runs several publishing businesses simultaneously and tests the AI tools that claim to help. The Practical AI is where honest findings go. No tech background, no PR relationships — just real tools tested under real conditions, written up clearly.

No affiliate links in this post. Full disclosure at disclosure