Knowledge Base

The discipline behind the decisions.

Frameworks, verdicts, trade-offs, and the operator-benchmark data that drives Articulate's positioning. Working in public. Every entry is dated, every claim is linked to a source, every "we accepted losing" is named. This is the spine — the Toolbox is the muscle, the Blog is the practice, the Clients page is the proof.

On this page
The Hype Radar — P/Q/S framework 16 hyped tools scored Operator stack patterns (8 global) UAE/GCC positioning gap Decisions log — last six Voice rules — always / never How we pick tools

The Hype Radar — P/Q/S framework

Every week, three or four "this will print money / replace your team / change your stack" posts cross the feed. Most are noise. A few materially affect what we ship. Without a framework, every one of them eats analysis time. With this framework, each gets a 90-second pre-score against three real-ground axes.

P
Personal productivity
Does it save hours per week on Articulate, GIG, or operator work? Score 0–3. "Might" doesn't count — would.
Q
Quality
Does output standard measurably rise — slop reduce, risk drop — on a real deliverable already in flight? Score 0–3.
S
Sellable
Can it slot inside an Articulate workflow we bill for — MEP, Design Agent, security audit, content engine, data-sovereign stack? Score 0–3.

The reusable rule. Sum all three for a total out of 9.

≥ 6 = act. Install, adopt, or build into Articulate.
4 or 5 = conditional. Re-score after a specific trigger fires.
≤ 3 = skip. Don't burn cycles. Quarterly re-score unless category materially shifts.

Hidden bonus rule — items that clear 6 only on P or Q are personal-use installs. Only items scoring meaningfully on S become Articulate product surface.

16 hyped tools, scored 16 May 2026

First-pass scoring of two hyped posts that crossed the feed this week — a "10 OSS repos that quietly print money" thread, and a "6 Claude Code plugins that replace an entire dev team" thread. Both reframed by the framework into install / conditional / skip.

Tool P Q S Total Verdict
n8n · open-source automation3238Act · Substrate for MEP delivery. Five reusable templates.
Superpowers · Claude plans before coding3317Act · Install. Stops jump-to-code on infra work.
Frontend Design · Anthropic plugin2327Act · Kills the AI-slop UI tell. Every client microsite benefits.
Security Review · Anthropic plugin2327Act · Run on VM gateway + every delivered codebase. Package as AED 2–4k line item.
Listmonk · self-hosted Mailchimp2125Conditional · Own email at scale. Selective resell to regulated UAE clients.
Plausible · privacy analytics1124Conditional · Self-host on articulate-ai.work. PDPL-clean pitch for regulated UAE/KSA.
Ghost · open-source Substack0123Skip · Re-score only if MEP adds managed-newsletter deliverable.
Cal.com1012Skip · Calendly works.
Coolify · self-hosted Vercel1012Skip · Re-score if hosting bills cross AED 500/mo.
Supabase · open-source Firebase1113Skip · Re-score only if pivoting to SaaS.
Code Review · Anthropic plugin1203Skip · Solo-operator. Re-score if collaborating.
Medusa · open-source Shopify0011Skip · UAE eCom is locked.
AppFlowy · open-source Notion0000Skip · Wrong sales cycle.
Penpot · open-source Figma0000Skip · Tiny UAE market.
Claude-Mem · persistent Claude memory0000Skip · Duplicates Cowork auto-memory + the CLAUDE.md vault.
gstack · 23 bundled role skills0000Skip · Duplicates Anthony's bespoke pm:* skills. Generic loses to fitted.

The hidden bundle

Four items combine into one coherent offer: n8n + Listmonk + Plausible + Security Review = a data-sovereign marketing stack. Sellable as a packaged Articulate sub-offer of the Marketing Engine Pilot for regulated UAE/KSA buyers — finance, healthcare, family-business modernisation. None of the other items combine like this. The decision to package it lives in the Decisions log.

Operator stack patterns — 8 operators publicly making money

Cross-reference against operators running productised AI/marketing services who openly publish revenue and stack. Eight benchmarked on 16 May 2026: Justin Welsh ($4.15M / 2024, ~86% margin), Pat Walls / Starter Story ($1.1M public, trending higher), Pieter Levels (~$3.1M ARR solo), Greg Isenberg / Late Checkout (agency + studio + media), Marc Lou ($1.032M in 2025), Daniel Vassallo / Small Bets (acquired by Gumroad $3.5M, July 2024), Nick Saraev / Maker School (~$250k/month, 10k+ members), Liam Ottley / Morningside AI.

Tools appearing in three or more of the eight stacks

ToolCountRole
Stripe6Payments + the public revenue artifact — screenshots, TrustMRR-verified badges
Notion4CRM + content + ops single source of truth
n8n / Make.com3Automation — n8n for enterprise tier, Make for solo
Cal.com / Calendly3Booking — Cal.com gaining share
Kajabi / Gumroad3Course + digital product checkout
Skool2Community + course hosting (replacing Circle/Kajabi for agency-adjacent operators)
Beehiiv2Newsletter — replacing ConvertKit/Mailchimp for creator-operators post-2023
Attio2Emerging default CRM for AI-native operators
X / YouTube8Distribution. Not optional.

Conspicuously absent across all eight: Salesforce, Marketo, HubSpot Marketing Hub, Pardot, traditional MAP stacks. The 2025–26 reference operator stack has standardised on Beehiiv + Attio + n8n + Stripe + Skool or Gumroad + Cal.com + X/YouTube. Articulate's Toolbox already overlaps on most of these; the gaps (Attio, Beehiiv, public Stripe artifact) are conscious choices logged in the Decisions log.

Three patterns we're borrowing

Public revenue artifact. Welsh-style "$623/month stack vs $125k/month revenue" page. Lou-style TrustMRR verified badge. Articulate will publish a verified pilot-outcome badge per client — verified email send volume, verified pipeline added — once the third install ships.

Mandatory tech-stack field on every case study. Pat Walls / Starter Story format. Every case study — Paul/BossCouple first, GIG next — lists the exact tools installed and the exact outcome. Buyer trust earned by specificity, not adjectives.

Productise one workflow harder, not five. Saraev, Lou, and Levels each have one repeatable thing they sell. The Marketing Engine Pilot foregrounds Workflow 01 — Inbound Triage + 5-minute WhatsApp response — as the hero installable. The other five workflows remain in scope but become "what else gets installed" not "the offer."

UAE / GCC positioning gap

Eight competitors surveyed: Traffic Digital (UAE, 100+ FTE), Tomsher (Dubai SME web shop), Digital Gravity (UAE, "AI-driven" decoration), Born Creators Group MENA (holding-co, four sub-brands), Voliom Automate (Dubai n8n + WhatsApp plumbing), Vista by Lara (Dubai Shopify-locked), AIST Agency (Riyadh, claims systems but no stack named), MAZ NEXA (Riyadh, HubSpot Solutions Partner).

What nobody is claiming

Solo operator + AI leverage as explicit value prop. Every competitor anchors on team size — Traffic 100+, Digital Gravity 200+, Born sub-brands. Nobody sells the inverse: fewer humans + better AI orchestration = faster, cheaper, more aligned delivery. Genuinely uncontested whitespace for a mid-market UAE buyer tired of account-manager churn.

Productised "Marketing Engine" as a fixed-scope pilot. Vista by Lara is the only shop with SKU-style offers — Shopify-locked, narrow. Nobody else sells a defined 60–90-day engine build at a fixed price. Everyone runs custom retainers or hourly.

Strategy + automation in one offer. Voliom = plumbing without strategy. AIST = strategy without named stack. Born = strategy fragmented across sub-brands. No one fuses GTM strategy, brand layer, and n8n/AI delivery under a single operator.

Stack whitespace

n8n-led delivery with strategy on top. Voliom owns n8n-as-plumbing; nobody owns n8n-as-marketing-OS. PDPL-clean self-hosted AI stack as a marketing headline. Voliom flags it; nobody else markets it. Federal Decree-Law 45 of 2021 + DIFC/ADGM data residency is a live buying concern for finance, healthcare, government-adjacent. Make it the headline, not the footnote. Anti-HubSpot positioning in KSA. MAZ NEXA's HubSpot lock-in opens room for an open-stack alternative — n8n + Postgres + Resend or Brevo + Cal.com — at 30–50% of TCO.

"The Marketing Engine Pilot — a productised, fixed-scope, AI-leveraged GTM engine on an open, PDPL-clean stack. One operator. Sixty days. AED 440k–660k year-1 outcome, not a retainer."
Nobody in the scan owns that exact sentence.

Decisions log — last six

Append-only. Every entry: what was decided, why, what we accepted losing. Full log lives in the project repo (~/ClaudeWork/Articulate/decisions.md). These six are the load-bearing ones.

2026-05-16
Four new public website surfaces — Clients, Projects, Knowledge Base, Blog
The Toolbox proved the long-form, prose, AI-readable single-file pattern. Four siblings extend it. Accepting we lose some marketing-site simplicity; mitigated by tight nav, every page anchored to the wedge.
2026-05-16
Productise one workflow harder, not five
The MEP foregrounds Workflow 01 — Inbound Triage as the named hero installable. Other five remain in scope but become "what else gets installed." Borrowing direct from Saraev/Lou/Levels in the operator benchmark. Accepting we lose optionality buyers who like menus.
2026-05-16
Borrow two artifacts from the operator benchmark
Stack-cost-vs-revenue page (Welsh-style). Mandatory tech-stack field on every case study (Walls-style). Both work in public, both cost ~nothing to maintain, both lower CAC. Accepting we lose: none. Stack-cost transparency is uncomfortable when costs spike — the page itself is the discipline.
2026-05-16
PDPL-clean stack becomes a marketing headline, not a footnote
Federal Decree-Law 45 of 2021 + DIFC DP Law + KSA PDPL compliance posture becomes a hero claim. Default stack (n8n + Postgres + Resend/Brevo + Cal.com self-hosted where required) is named explicitly as data-sovereign. Accepting we lose: buyers who don't care about data residency. They aren't the target.
2026-05-16
Solo operator + AI leverage is the explicit Articulate wedge
The hero message names what nobody in the UAE/GCC competitor set claims: fewer humans + better AI = faster, cheaper, more aligned than a 100-person agency. Genuinely uncontested per today's competitive scan. Accepting we lose: buyers who read "solo" as risk — handled by case studies, handover docs, and the day-31 fire-the-consultant clause.
2026-05-16
Hype Radar P/Q/S framework adopted
Every hyped product, repo, plugin, or launch gets scored P/Q/S before deeper evaluation. ≥ 6 = act. 4–5 = conditional. ≤ 3 = skip. Source-of-truth doc at ~/ClaudeWork/Claude/hype-radar.md. Public sibling on this page. Accepting we lose: the dopamine of chasing every new tool. Quarterly re-score is the safety net.

Voice rules — always and never

Articulate AI's voice is direct, builder-not-agency, British, short-sentence. Confident, dry, occasionally blunt. The rules below apply to every page on this site, every DM, every proposal, every case study. They're enforced in review.

Always

  • Short sentences. Active voice.
  • AED for money, not USD or GBP unless context demands.
  • UAE-specific examples — WhatsApp-first, broker commissions, off-plan launches, Ramadan and summer cycles.
  • Numbers where possible. Specifics over abstractions.
  • Frame failures the buyer feels: "leads go cold," "dormant CRM," "you reply Monday morning."
  • Builder framing — install, hand over, own it.
  • Challenge weak ideas, including the buyer's.

Never

  • "In today's fast-paced world."
  • "Synergy," "leverage," "ecosystem," "robust," "seamless," "best-in-class."
  • "Empower," "unlock," "transform" as buzzwords.
  • Emojis (unless the buyer uses one first).
  • Long agency-speak intros before getting to the point.
  • Generic case-study fluff — every number is real or framed as "expected."
  • Multi-page maze sites or pitch decks. One page that converts.

Before shipping any page or post or DM, the draft has to answer four questions. Does the buyer feel a specific failure in the first sentence, or am I describing benefits? Did I say "install / hand over / own" anywhere, or did I slip into "we partner with you to..."? Are the numbers real, expected, or made up? Would Anthony say this out loud to a Dubai broker over WhatsApp? Any wrong answer — rewrite.

How we pick tools

Three tests, in order. Every tool earns its place by passing all three. The full live working list is on the Toolbox page — this section is the rule.

  1. 01 Does it do the job in one tool, or three? One beats three.
  2. 02 Does it run on the client's stack after handover, or lock them into us? Hand-over beats lock-in.
  3. 03 Is it best in category right now, or just the most familiar? Best beats familiar.

Test three is the one most operators fail. Familiarity is comfortable. The category moves every quarter. Our review cadence — monthly observation, mid-month audit, quarterly re-pilot, annual zero-base — is the discipline that keeps test three honest. Full cadence detail on the Toolbox.


The frameworks live on this page. The output lives in the work.
Last updated · 16 May 2026
Reviewed weekly · written by Anthony