Skip to main content

Reviewers and press

Test it properly, and say what you find

If you are writing about cpywrk, this page has the facts you need, a test plan designed to break it rather than flatter it, and access with nothing attached to it. Everything below is current as of 2 September 2026.

Read this first

Nothing here is conditional on what you write

Reviewer access is not a trade. Putting it in writing so there is no ambiguity later:

  • No payment, in either direction. We do not pay for coverage and we do not buy placements that are presented as editorial.

  • No requested rating, link or anchor text. Access is not conditioned on a score, a followed link, a particular phrase, or a link at all.

  • No copy approval. We will correct a factual error if you ask us to check one. We will not ask to see the piece first.

  • One ask, and it is not about tone. Please disclose that complimentary temporary access was provided. That is standard practice and in several jurisdictions it is required.

If the product does not do what this site says, that is the more useful article, and we would rather read it than argue with it.

The facts, in one place

These are the canonical figures. Where this page and any other page disagree, this page is the error and we want to know about it.

What it is
A content pipeline. It takes a topic through keyword research, an approved outline, drafting, six quality checks and publishing. It is not a chat-style writing assistant.
Who runs it
Founder-led, based in Denmark. Company registration is in progress and the contracting entity will be published on the legal pages when it completes.
Plans
Trial (free, one full article, no time limit), Starter €199/mo, Growth €499/mo, Managed from €999/mo. Annual equivalents are €159 and €399. Managed is sold after a conversation, not through checkout.
Volume
Starter 13 articles a month, Growth 35. Managed volume is agreed with each customer under fair use and is deliberately not published, because it is a commercial term rather than a plan feature.
Publishing
Direct to WordPress and Shopify. Anything else goes out through a webhook. LinkedIn, X and Instagram versions are generated from the same brief but are never posted for you.
Quality
Six checks: search, AI answers, brand voice, brief compliance, machine tells, and your past corrections. The method is published in full, including what each one does not prove.
Languages
The product runs end to end in English, Dutch and Czech. The marketing site is localised for all three; the blog and legal pages are English.
AI provider
Anthropic. Customer content is not used to train models.
Approval
On the self-serve plans nothing publishes until the customer approves it. On Managed, cpywrk approves against a brand brief the customer signs at intake. The two are deliberately different products.

What reviewer access includes

30 days, no card

No card at signup and no auto-renew at the end. It expires rather than converting.

Five full articles, one business

Every cpywrk account is one business, so all five run on the same one. Enough to test the system rather than judge one polished output. The protocol below explains what each article is for.

Every capability that is genuinely live

Whatever is shipped on the day you start, including the publishing integrations. Nothing is switched on specially for a review.

The underlying evidence

Downloadable briefs, drafts, check results, your corrections, and the final outputs. You can inspect the process, not just the product.

A 20-minute founder setup

Optional. Some reviewers would rather start cold to see what onboarding is actually like, which is a fair test and we will stay out of the way.

A technical support route

For when something breaks during the test. If it breaks, that belongs in the review too.

A five-article test protocol

Use it, ignore it, or break it. It exists because the claims worth checking are about correction learning and repeatability, and neither shows up in a single article. Each row names the question it is designed to answer.

Five-article reviewer test protocol
# What it tests The question it answers
1 Setup and baseline The initial brief, the settings you chose, timestamps, and the first draft. Does setup produce a usable, on-brief baseline?
2 Correction learning The original passage, your correction, and the regenerated output. Is the correction retained, and can you see that it was?
3 Repeatability A second topic for the same business, and a correction you turn into a standing rule. Does it stay consistent without becoming repetitive?
4 Standing rules The rule you set after correcting article 3, and the passage in article 4 where it is applied without being repeated. Does a correction become a rule, or do you make it again every time?
5 Publish and export Final check results, the export or publish log, and the rendered page. Does the whole workflow work outside a demo?

Articles 2 to 4 are the ones that matter. Correction learning and repeatability are the two claims that separate a pipeline from a text box, and they are the two a one-article demo cannot show.

Before you ask for access

There is already output to judge. Every article on the blog was produced by the pipeline described here and passed the same six checks. Nothing on it was hand-written for the website, which is the point of publishing it.

Getting in touch

Press enquiries, reviewer access, a founder interview, or a factual correction to something already published. Same address, and it reaches a person rather than a queue.

customer.success@cpywrk.com

Worth saying what you are working on and roughly when you need it. If a claim on this site does not hold up, say that too and we will fix the page.

Questions

Does cpywrk require a positive review in exchange for reviewer access?
No. cpywrk attaches no condition to reviewer access: no required rating, no link, no anchor text, no approval of the draft before publication, and no expectation of a positive tone. Critical coverage is welcome. The only request is that reviewers disclose that complimentary temporary access was provided, which is standard practice and required in several jurisdictions.
Why does cpywrk offer five articles to reviewers rather than one?
One article can show a demo; five can test a system. The claims worth checking are about correction learning and repeatability, and neither is visible in a single output. Every cpywrk account is one business, so the five articles run on one account, which is the smallest test that exercises both claims without becoming an open-ended free account.