hey squad
shall we settle in?

01 · SEO · 03/06

AI writing.

Not mass content. Not copy-pasted ChatGPT. Briefed content that actually answers.

You lay 40 % brief, 30 % Claude-assisted generation, 30 % human QA. Your subject-matter experts review in 30 minutes. The produced content sounds like you, not like a tool. Volume held without editorial dilution.

By Camille · Voice & content

Get my production quotedSee the AVD Revival mission

02 · OPERATION NOTE · OPS AI WRITING

AI writing, decoded

The field report

You have a framed pillar-satellitea core page + its linked sub-pages plan, but your editorial team can't keep up. Or you pay a provider 800 € per article that sounds generic. You're looking to scale volume without breaking the brand voice. Often the moment.

The cost of waiting

Without a briefed AI pipelinethe serial production chain, you choose between volume and quality. You publish little but well, or much but generic. The shortfall is double: you lose the volume traffic, and the credibility that quality brings. Both end up costing you.

In black and white

A pipelinethe serial production chain that produces 4 to 8 satellitesthe sub-pages linked to the pillar page per month, briefed and reviewed. A brand voice preserved at 90 % across the entire set. A cost per article divided by 2 to 3 vs traditional writer, at equal or better quality.

The manoeuvre

We start by framing your brand voice, before generating a line. Detailed brief per satellite (intent, outline, sources), generation assisted by Claude APIthe Anthropic AI model, 30-minute human review per text. We work with your editorial team, never replacing it.

The lock that breaks

A production volume that follows your editorial plan without diluting the voice. A topic coverage built month after month. A clean base to amplify on local SEO and newsletter behind.

03 · WHAT WE WON'T WRITE IN AN RFP

AI writing is an editorial leverage tool. Not a writer replacement.

Briefed AI is only useful if you keep the human brief upstream and the human review (quality assured by a human) downstream. Otherwise, it's fast content that sounds generic and Google demotes via Helpful Content Updates (Google updates on content quality). The right question before signing: who on your side will brief and review the AI texts? If the answer is fuzzy, the right order is brand voice and pillar pages first, then plug in AI. The broader agency vs freelance vs in-house debate sits here.

  • 01

    Held production volume

    4 to 8 satellites per month, briefed and reviewed. Your editorial plan moves forward without cracking. The pace holds over 12 months without exhausting your editorial team.

  • 02

    Preserved brand voice

    Your AI texts sound like you at 90 %. Detailed brief + reference examples + expert review hold the voice. Not raw ChatGPT, but piloted Claude.

  • 03

    Cost divided by 2 to 3

    A briefed AI satellite costs 250-500 € HTVA. A senior writer satellite costs 600-1,200 €. At equivalent quality, the yearly gap is significant on a 50-article set.

  • 04

    Reproducible pipeline

    Your briefs and reference examples stay yours. If you change provider tomorrow, the AI pipeline keeps running. No dependency on someone walking away with their method.

04 · THE MISSIONS THAT PROVED IT

We deployed it on these grounds.

AVD Revival, 250 GTO reinterpreted by Björn Arens

AVD Revival

High-end automotive · revival

6-month Claude-briefed AI writing pipeline.

Read the AVD Revival mission
Ecostal Group, B2B photovoltaic e-commerce, 3 storefronts EBE/EFR/YMT

Ecostal Group

B2B photovoltaic e-commerce · 3 storefronts

Briefed multi-market production over 18 months.

Read the Ecostal mission
Brasero estate, corporate retreats and weddings in Liège province

Brasero Group

Hospitality & events · 2 sites in Liège

Briefed writing across territories over 24 months.

Read the Brasero mission

05 · THE PLAY-BY-PLAY

Four steps. Four weeks for the pipeline. Then it runs.

  1. STEP 01

    We frame your brand voice.

    Audit of your existing corpus: 10-20 texts representative of your tone, vocabulary, structure. Invariants extraction. Construction of a system brief (instructions given to the AI) and 5-8 reference examples that will run through Claude Sonnet (opens in a new window) (fast version) or Claude Opus (powerful version). Validated by you before generating a single satellite.

  2. STEP 02

    We freeze the brief per satellite.

    For each satellite: intent, headings and subheadings outline, target query, sources to cite, length, specific tone. 600-1,000 words brief. Client sign-off before generation. No silent scope drift, no blind generation.

  3. STEP 03

    We generate then review.

    Generation via Claude Sonnet for volume or Claude Opus for complex texts, initial pass 80 % quality. Human review 30 minutes per text: factual accuracy, voice, flow, sources. 1-2 iterations max to reach 95 %. No text ships without human review.

  4. STEP 04

    We document and hand over.

    Full pipeline documented in Notion: system briefs, reference examples, brief templates, review checklist. Handover to your editorial team or existing agency in 1 week. Positions and impressions tracking on target queries in Google Search Console. Total ownership.

06 · THE FLOW AT A GLANCE

The path each text takes between the human brief, Claude and publication.

PRODUCTION CHAIN FOR ONE TEXTAUDIENCEINTENTSOURCESOUTLINECLAUDE SONNET/ OPUSinitial pass 80 %FACTSVOICEFLOWSOURCES30 MIN/ reviewed textBRIEFprecise · human40%CLAUDEassisted generation30%HUMAN QAreview · 30 min30%T = 0T = publish95 % quality1-2 iterations max30 min · piecenot 6h writing÷ 2-3 cost cut€250 vs €600-1,200Reusable pipelinebriefs + few-shots = ownership"the right word is worth €30 of Meta" · Camille · content & SEO

The human brief weighs 40 %, Claude generation 30 %, human review 30 %. That distribution holds the voice and dodges Helpful Content Updates.

07 · NOT YET FOR YOU IF

Three cases where briefed AI writing is not the top priority.

  • You have no framed pillar-satellite plan yet.

    Plugging AI on a site without a pillar or satellite plan is producing fast content that does not rank. The right order is plan framing first, then briefed AI scaling. Otherwise we multiply editorial noise.

  • You publish less than 4 articles per month.

    The AI pipeline framing effort (3-4 weeks) does not pay back below this threshold. At this pace, a human writer stays simpler and just as cost-effective. The AI lever makes sense from 4 monthly satellites onwards.

  • No one on your side will review the generated texts.

    An un-reviewed Claude text sounds generic in six weeks and Google demotes it via Helpful Content Updates. If no internal resource is planned for human review (30 minutes per text), the cost will not be paid back by simple volume gain.

08 · THE QUESTIONS WE ACTUALLY HEAR

Questions whispered after the second meeting. Honest answers.

On the Belgian SME market in 2026, traditional writers cost 600-1,200 € per satellite, with 4-6 monthly articles max capacity. The annual gap on a 50-satellite set runs between 25,000 and 40,000 € HTVA. The real cost is not the lost money, it is the editorial pace that does not follow and authority that builds twice as slowly. The broader agency vs freelance vs in-house debate sits here.

Not the same. Claude Sonnet (Anthropic's fast version) and Claude Opus (powerful version) hold brand voice better on briefed long format, keep syntactic nuances in formal French, and invent fewer fake technical sources. ChatGPT (OpenAI model) stays strong on conversation and short variations. For briefed long-form SEO writing, Claude wins on initial pass (80 % vs 65 % for ChatGPT). The lever pays off most when a pillar-satellite plan is framed behind.

Not for the AI nature itself, as of late 2026. Google clarified in 2024 that what matters is value, not origin (see Google Search Central (opens in a new window)). The Helpful Content Updates 2024 and 2025 (Google updates on content quality) however massively demoted sites publishing raw un-reviewed ChatGPT with no angle. The rule that holds: briefed AI + human review = OK. Un-reviewed chain AI = almost certain algorithmic penalty. Mandatory infrastructure prerequisite: clean technical SEO behind.

The briefed AI pipeline does not change the per-article yield (equal or better quality). It changes the pace: 4-8 monthly satellites instead of 2-4. Over 12 months, it doubles topic coverage and incremental SEO traffic. On a framed plan, we observe +60 to +150 % SEO traffic on target queries vs writer-only production. Editorial consistency holds on a brand voice framed upstream, discovery by answer engines on AI search (GEO) downstream.

Claude API side (Anthropic model): Sonnet costs 5-15 € per generated satellite, Opus 15-40 € depending on length. Internal human review side: 30 minutes per text, 1 day per month for 4 satellites. If we leave, you keep the Notion system briefs + reference examples + brief templates + relay procedure in 1 week. The equipped AI stack also runs on workflow automation.

Yes. We do this on 60 % of our AI missions. The rule: who steers the brief and the review. If your writers become AI supervisors, we train them in 2 weeks on the Claude pipeline. If we generate and they review, we frame the review checklist. We never step on the other agency without saying so.

Field note

AI writing is not a robot taking the pen instead of you. It is a tool that scales a well-briefed human. The right word is worth 30 € of Meta, but someone still has to know which word to lay down, and where. We take almost as much time on the brief as on the review. That is where the voice catches, like a melody on a score. Claude plays the notes. It does not invent them.

Camille · Voice & content
CamilleVoice & content · HeySquad

Other sub-services SEO & CONTENT

We look at your editorial plan. We tell you what briefed AI could do for you.

Get my AI pipeline quotedBack to the SEO & content service