What AI Writing Instruments Get Flawed (And The Stack I Use As a substitute)

AI writing instruments make the writing half sooner, however writing was by no means the laborious half.

The laborious half in content material advertising is the knowledge—concepts, verified information, and reference materials. And that’s precisely the place these instruments fall brief.

I realized this after producing 40 articles by means of Claude. I’d tried the writing instruments first, however they only couldn’t deal with the half that truly issues. And by “AI writing instruments” I imply the platforms constructed on high of LLMs—Jasper, Frase, Writesonic, that class. What I used as an alternative was the LLM straight, with my very own information and course of round it.

On this article, I’m sharing the 5 issues I bumped into and the way I deal with them now.

I’m not naming the particular instruments I examined. They’re not dangerous merchandise. When you don’t have robust writing or search engine marketing abilities, otherwise you don’t have time for a extra hands-on course of, they’re a effective selection. That content material is best than no content material. However you probably have the talents and wish to push high quality, they change into the ceiling, not the ground.

1. The analysis drawback: AI “analysis” simply recycles no matter’s already rating 

Most AI writing instruments “fact-check” the content material they generate by cross-referencing it in opposition to no matter ranks on Google.  Competitor advertising pages. Outdated weblog posts. Articles that copied their information from different articles. In observe, they’re laundering errors by means of consensus—if three improper sources agree, the AI treats it as truth.

And that’s a straight path to worldwide meta-spam.

I imply, once I let writing instruments deal with analysis, I acquired improper costs, incorrect options, and database numbers off by thousands and thousands. More often than not, it simply pulled from biased sources and had no option to know they had been dangerous. 

One of many instruments I examined used Gemini Deep Analysis because the article foundation. However Gemini—and I believe each different AI assistant—does the identical factor.

Once I wrote a comparability masking eight merchandise, I wanted eight separate fact-checked paperwork, one per product, plus a method information, an enhancing guidelines, and a immediate with required components. That’s 15-20 information I wanted the AI to reference all through the method. No writing device I examined might deal with that.

My resolution: at all times construct your individual reference information 

Construct verified information information for each product and competitor you cowl. Begin with a information base on your personal merchandise, in a type the place you may simply generate paperwork from it: pricing, options, use circumstances, all the important thing numbers. I really vibecoded a device for that.

If it’s essential characteristic rivals in your content material, put together paperwork for the components you need referenced: their pricing pages, characteristic lists, limitations, and many others. I downloaded competitor touchdown pages, took screenshots, and vibe-coded a scraper to drag pricing and options from official sources.

By no means begin any AI content material venture till your information information are carried out. In case your venture is supposed to take 4 weeks, use three weeks for these information.

2. The method drawback: writing instruments attempt to one-shot your article, however good writing doesn’t work that method 

Writing instruments are meeting traces: configure inputs, press generate, gather output. However writing is nearer to cooking—you style at each stage, add some unplanned components, or perhaps flip the factor into one thing else. 

It doesn’t matter how a writing device handles model voice. Whether or not it’s a dropdown, a method file, or a set of directions, the outcome at all times wants enhancing. Getting our voice proper took 5 – 6 rounds per article. I’d learn a draft again and say “that appears like a press launch” or “put the quantity first, you’re burying the lead.” You want a dialog for that.

That is additionally an interface drawback. Enhancing AI-generated textual content means working at each stage: rewriting a single sentence, restructuring an entire part, fixing a sample throughout the whole article. In a chatbot, I simply requested for what I needed in plain English. Writing instruments gave me mounted enhancing choices that couldn’t deal with that vary.

My resolution: break your course of into repeatable prompts or abilities 

Break your workflow into repeatable duties and develop prompts for every:

• Truth-checking.
• Inner consistency checking.
• Type and construction enforcement.
• Product positioning enforcement.

Trial and error till every immediate nails it.

In a while, these prompts can change into your Claude abilities, if/while you determine to make use of automated content material workflows.

Tip: For crucial steps, I ran my prompts twice, or ran the identical verify by means of a second AI to catch something the primary one missed.

3. The size drawback: writing instruments deal with each article as an island 

Writing instruments encourage you to consider automating content material at scale. Some even supply workflow options for it. However I discovered them irritating in observe: laborious to construct, human-in-the-loop performance could be very restricted, and the output drifts the extra nuanced your necessities get. 

AI assistant already solved this, and Claude Code took it to the subsequent stage. I might sort “scan each article for Product X’s pricing and verify it in opposition to the reference file” and it could do it. When one thing wanted adjusting, I simply informed it.

That’s performance that writing instruments don’t supply, although the underlying LLM is able to it.

My resolution: get used to working with Claude Code 

In Claude Code and OpenAI Codex, one instruction kicks off the entire course of. Tt fetches search engine marketing information, pulls from my reference information, grabs what it wants from the net, and writes the article in phases. I outlined the phases, then let it run whereas I did one thing else. 

That is additionally the place analysis instruments plug in. MCP integrations like Ahrefs’ allow you to pipe actual information straight into these workflows—we’re experimenting with a full Claude Code pipeline the place search engine marketing analysis occurs mechanically. In case your device doesn’t assist MCP but, pull the info manually. Even screenshots work, so long as you give the AI particular information to work on.

4. The economics drawback: when the wrapper prices greater than the engine 

A chatbot subscription prices $20 a month and offers you the most recent mannequin with no article or phrase limits. The writing instruments I examined price $50-200 a month, one even $2k a month, and ran older fashions with caps on how a lot you would generate. Appears like paying extra for much less.

Right here’s an instance. To jot down one of many articles for the experiment, I pulled the top-cited articles for my key phrase (utilizing Ahrefs’ Model Radar), then had Claude undergo these pages to extract the construction and use that as an overview template for content material era. Then I requested it to weave in my very own concepts. Analysis, construction, writing—multi functional dialog, controlling each stage.

However perhaps I’m improper. Perhaps a writing device with the whole lot on board is extra your fashion. I’ll go away it to you to determine what makes extra sense economically. I don’t wish to inform you what to do together with your cash, however I do know that for my wants, I’m by no means going again to AI writing instruments.

There’s additionally one thing a bit self-defeating in regards to the AI device ecosystem. Each time an LLM supplier releases a greater mannequin, lots of the instruments constructed on high of it lose a part of their motive to exist.

My resolution: make investments extra in what you feed the AI

• Analysis instruments that go deep. Wealthy key phrase information, search intent evaluation, aggressive gaps, AI-preferred content material codecs, and many others. Writing instruments bolt on a surface-level model of this. Devoted platforms have years of infrastructure behind them (right here’s ours).
• Your editorial system. Immediate libraries, fact-checking workflows, fashion enforcement, Claude or Codex abilities. The stuff that retains your judgment within the loop at each stage. Similar precept because the reference information: spend money on the inputs.

This setup additionally makes it simpler to adapt when fashions change or your content material wants shift. It’ll click on after the subsequent part.

5. The content material technique drawback: writing instruments provide you with one course of for 2 very totally different jobs 

Writing instruments assume all content material works the identical method. Feed it a key phrase, get an article. However I see content material splitting into two tracks in our line of labor, and writing instruments can’t deal with both one correctly.

The primary is searchable content material. Product documentation, assist articles, comparability pages—the stuff most groups handled as a chore. It’s all of a sudden essential as a result of if an AI mannequin can’t floor its reply in one thing you revealed, it’ll use no matter it finds. Or hallucinate. Your product documentation is your model’s voice inside each AI dialog now.

Right here’s what that appears like when it really works. I requested AI Mode, “What number of manufacturers are you able to observe in Model Radar?”, and it cited our docs straight.

And right here’s what occurs when there’s a spot: no official supply cited. Fortunately, the truth that I requested AI mode about acquired talked about in one other piece, however that was nearly accidentally.

The second, I believe, is shareable content material. Really human-first content material. Stuff that comes from private expertise and may’t be templated. My AI misinformation experiment is an instance: it ranked for nothing, however drove 24k visits and extra social traction than I might depend.

My resolution: select flexibility over comfort

You may chat with the preferred AI assistants to identify holes, or arrange monitoring in a device like Ahrefs Model Radar to do it at scale.

For shareable content material, construct an concept pipeline. Begin a scrapbook. Retailer concepts, information, quotes, social posts, e-newsletter excerpts, and something you may wish to give AI entry to later.

You should use Notion, Evernote, no matter fits you. However think about vibecoding a customized device, like my colleague Louise. That method, you may bake in options like an “instance finder” that surfaces related assist for claims in your writing, or simply generates content material concepts out of your materials on the spot.

One other concept: arrange an AI agent that scours the net for content material concepts on a schedule. I constructed one with Relay that goes by means of LinkedIn and Reddit conversations (honest use) each 7 days. It helped me keep on high of all the brand new content material popping out sooner than ever and keep sane.

If you wish to preserve a fixed pulse on new content material in your house, strive our new device, Firehose. It streams the net in actual time on any subject you outline, with superior filtering. You describe what you’re on the lookout for in pure language, and it’s able to go. You can too join it to your AI brokers by means of the API.

Remaining ideas

Thanks for studying! If in case you have any questions or feedback, let me know on LinkedIn.