Skip to main content

Content engine

Drafts that sound like you, every time.

The MarquIQ content engine generates platform-native drafts, enforces your voice with real examples, strips AI tells with a deterministic scrubber, and runs a second critique pass before anything queues.

Writing one post is easy. Writing platform-native variants across 26 surfaces every week, without sounding like a content farm, is the real job. The content engine is the part of MarquIQ that does that job.

What you get

  • Voice enforcement that actually works

    Paste three to five real posts. The model pattern-matches to your tone. Every draft reads like yours, not like a model default.

  • Scrubber + critique pass

    A deterministic scrubber strips em dashes, leverage, robust, seamless, and other tells. A second LLM pass rewrites anything still generic.

  • Platform-native variants

    One research brief produces a draft per connected surface across 26 platforms. Each variant is structured, lengthed, and toned for its platform, not a retrofitted template.

  • Banned phrases per product

    Ship with a sensible default list. Add your own taboo phrases with regex support. Violations never reach the approval queue.

  • Cross-product dedupe

    Jaccard similarity over a 14-day window prevents you from posting the same idea twice across products. Dedupe windows are configurable.

  • Evergreen recycling

    Posts older than 90 days that performed well automatically re-enter the queue with a rewrite, never the same text.

How it works

  1. Step 1

    Ingest the product

    MarquIQ crawls your product URL, ingests your messaging, and builds a fact sheet the writer can cite.

  2. Step 2

    Generate with voice examples

    The first prompt pattern-matches to your pasted voice samples and produces a platform-native draft for every connected surface (up to 26).

  3. Step 3

    Scrub the obvious tells

    Deterministic replacement of em dashes, leverage, robust, seamless, navigate, and other known AI markers. Always runs.

  4. Step 4

    Critique and rewrite

    A second LLM call reads the scrubbed draft and rewrites anything that still reads generic. This is the step most tools skip.

  5. Step 5

    Approve or auto-send

    You review on mobile or desktop. Approve, edit, or reject in one tap. High-confidence drafts can auto-send on low-risk surfaces.

Frequently asked questions

Does the content engine write platform-native variants?

Yes. One research brief produces platform-native drafts for every connected surface across 26 platforms: X, LinkedIn, Reddit, Bluesky, Mastodon, Threads, Dev.to, Hashnode, Medium, Indie Hackers, Hacker News, Product Hunt, Substack, Ghost, Quora, Stack Overflow blog, YouTube community, TikTok captions, Lemmy, Pinterest, Tumblr, Beehiiv, ConvertKit, newsletter paragraphs, Featured.com pitches, and HARO-style expert answers. Each variant is structured and toned for its surface, not a template.

How does voice enforcement work?

You paste three to five real posts in your voice. The first prompt pattern-matches to those examples. A deterministic scrubber then strips em dashes and known AI tells. A second LLM critique pass rewrites anything still generic.

Can I ban specific phrases?

Yes. Per-product banned-phrase lists with regex support. Common starters are automatically on: em dashes, leverage, robust, seamless, navigate (metaphorical), in today's fast-paced landscape.

Which AI model powers drafting?

OpenAI GPT-4o is the default for generation, GPT-4o-mini for classification. The provider layer is pluggable so the model can swap without changing your configuration.

See the content engine in production.

Plug in your product. See platform-native drafts for every connected surface in under two minutes. No credit card up front. Fourteen-day trial.

Start the trial

Related reading