Step 3 · Write
AI content creation
Pages that sound like you, hold up on facts and can be cited by AI assistants.
Traze writes the pages in your content calendar: articles, buying guides, service pages, comparisons and FAQs. Every page starts from a brief and your brand profile, is checked against facts and sources, gets a citability score and waits for your review before it enters your CMS. Fast enough for dozens of pages a month, careful enough to publish under your own name.
What you get · concretely
What AI content creation with Traze includes
- 01
Brand profile
Tone of voice, terminology, products, audiences, allowed and forbidden claims, sample texts. Built from your site and your input, and the foundation for every page.
- 02
Writing from the brief
Every page follows the brief from your strategy: target question, sub-questions, entities, sources, structure and internal links. Not a loose prompt, but an assignment with context.
- 03
Fact and source checking
Numbers, claims and product details are checked against your own data and the designated sources. Uncertain passages are flagged for the reviewer, not smoothed over.
- 04
Citability score per page
Direct answers, self-contained passages, entity coverage, structure, schema and source attribution, scored before the page goes live, with pointers per section.
- 05
Metadata, schema and internal links
Title, description, headings, FAQ and Article schema and contextual links to hubs and related pages come as standard.
- 06
Human review
Every page lands in a review queue with its score, sources and flagged passages. You or your team approve, edit or send it back. By default nothing goes live without approval.
What it is
What AI content creation at Traze is
AI content creation at Traze is not 'generate a blog post about X'. It is a production process with fixed steps: a brief from the strategy, a brand profile that sets the voice and the boundaries, a writing pass, a check on facts and sources, a citability score and a human review. The language model is one of the steps, not the process. For data-driven pages, such as a page per location or product, the same process runs in batches. See how this works at scale.
The brand profile is the foundation. Traze reads your existing site, product copy and any style guide and records how you sound: formal or informal, which words you use and avoid, which claims you can and cannot make, which competitors you do not name. You correct that profile and it grows with every review. For multilingual brands Traze builds a profile per language, so an English page is not a translated Dutch page. Read how that works per language and market.
Every page is built for two readers: the visitor who wants an answer, and the AI assistant looking for a source to cite. That is why every section opens with a direct answer, facts sit in self-contained passages, entities are named explicitly and schema sits underneath. Read how Traze measures whether you get cited.
And every page has a place in the whole: it comes from a cluster, links to the hub and its sibling pages, and fills a gap the strategy identified. See how the strategy sets the briefs.
How it works · step by step
How a page comes to be
Step 1
Pick up the brief
Traze takes the next page from your calendar: target question, sub-questions, entities, sources, intended format and internal links. For product pages your own data comes along.
Step 2
Research
Traze reads the sources in the brief, your own data and what currently ranks and gets cited for this question, and works out what a better page needs to contain than what already exists.
Step 3
Write from your brand profile
The page is written in your voice and within your boundaries. The structure follows the brief: direct answer at the top, an H2 per sub-question, tables and lists where they help, FAQ at the bottom.
Step 4
Verify
Numbers and claims are compared with the sources and your data. Anything that cannot be verified is flagged. Traze also checks for repetition, filler and passages that resemble existing pages too closely.
Step 5
Score
The citability score assesses direct answers, self-contained passages, entity coverage, structure, schema and sources. Below the threshold, the page goes back for a rewrite pass, with pointers per section.
Step 6
Review and publish
The page appears in your review queue with its score, sources and flags. After approval it goes into your CMS with metadata, schema and internal links, and is submitted for indexing.
Why it matters
Why this beats a writer alone or a model alone
A good writer delivers a few strong pages a week and knows your brand after a few months. A bare language model delivers a hundred pages an hour that all sound the same and whose facts nobody has checked. Neither is what you need to claim a topic in Google and in AI answers. Check for free whether AI assistants already mention your brand.
Traze combines the pace of the model with the oversight an editorial team provides: a profile that holds the voice, a check that tests claims, a score that judges the page the way an AI engine does, and a human who has the final say. You get volume without your brand getting diluted. See what each plan includes.
And because every page is measured after publishing, the process learns. Pages that get cited and convert feed the profile and the briefs; pages that lag get rewritten. See how Traze tracks performance per page.
- citability score per page, before publishing
- 1
- of pages arrive in the review queue with sources and flags
- 100%
- pages per month, depending on your plan
- 10–40
Comparison · old way versus Traze
Standalone AI tools or writers alone versus Traze
| Area | Standalone AI tools or writers alone | Traze |
|---|---|---|
| Starting point | A prompt or a title | A brief from the strategy: target question, entities, sources, structure |
| Voice | Generic, or dependent on whoever happens to write | Brand profile that steers every page and grows with your reviews |
| Facts | Unchecked, or only if the writer happens to notice | Checked against sources and your data; uncertain passages flagged |
| Citability | Unknown until you go looking in AI answers yourself | Score before publishing, with pointers per section |
| Review | Paste into your CMS and wait and see | Review queue with score and sources; approve, edit or send back |
| Publishing | Manual, with metadata and schema handled separately | Straight into your CMS, with metadata, schema and internal links |
| Learning | Every page starts from zero | Results per page feed the profile and the next briefs |
Use cases · where it works
Where AI content creation pays off most
Long-tail articles and FAQs
The hundreds of specific questions your customers ask that nobody on your team has time for. Each one a direct answer that Google and AI assistants can use.
Category intros and product copy
Unique text per category and product from your feed, in your voice, without every page sounding the same.
Service pages and location pages
Conversion-focused pages per service and service area, with the local questions and objections that come up in your sales process.
Comparisons and buying guides
'X vs Y' and 'best X for …' pages with honest trade-offs. Exactly the format AI assistants cite when someone asks which to choose.
Refreshes of existing content
Pages that have aged or sit below the score are rewritten while keeping what works, including updated facts and schema.
Honestly · risks and limits
Will Google penalise AI content? And other risks
The honest question first: no, Google does not penalise content because it was written by AI. Google's guidance has said since early 2023 that the production method does not matter, quality does. What Google does penalise is scaled content abuse: producing many pages with the main goal of manipulating rankings, without value for visitors. Whether that happens with AI, with cheap writers or with copied text. Since the spam policy update of March 2024 this is actively enforced.
So the risk is not in the tool, but in what you do with it. Pages without insight of their own, with repeated paragraphs, made-up numbers or a voice that comes from nowhere are ignored by visitors and weighted down by Google's quality systems. The same goes for AI engines: they cite sources that are specific, verifiable and consistent.
Traze builds in the safeguards against that. Every page has a brief with a reason to exist. The brand profile prevents generic text. The fact check flags what is wrong or cannot be verified. The citability score holds back pages that are not good enough. And the review means a person takes responsibility for what appears under your name.
What we do not promise: that every page ranks or gets cited immediately. Content is one factor, alongside authority, technical health and time. What we do promise: that you see exactly what was written, why, based on which sources, and that you can always edit or roll it back.
Try it yourself · free
See it at work on your own site
This free tool shows part of this capability. No login, instant results.
Content Scorer
Content Scorer
Paste a URL and a keyword. Within three seconds you get a citability score for AI engines, with the concrete change per component.
Try:
Results · example data
What it looks like when Traze does the work.
Vloerenwinkel
Flooring
+30,000
visitors / month
AI citations: 0 → 210
Edelstenenhandel
Gemstones
+12,700
visitors / month
AI citations: 12 → 153
Wasruimtemeubels
Laundry-room furniture
+8,259
visitors / month
AI citations: 0 → 142
Illustrative numbers to show how the system works — not a real customer.
Questions · about AI content creation
Frequently asked questions
Get started · self-serve or with a demo
Let Traze run this for you
Start on the Start plan yourself, or book a demo where we show this capability on your own site. Pricing is public, with a rolling monthly option.