Closed-loop AI systems

Systems that run
without being watched.

We build AI systems that check their own work and correct themselves. One of them writes and publishes your content every day, and you can buy it for EUR 129 a month. The other kind we build inside your company.

See both →Talk to us →
Autopilot · EUR 129 / moAI Lab · by engagementBuilt in Stockholm
Loading Ari...
Chapter I
How the loop works

Six steps the system runs every single day.

§ 01Diagnose

We start with where you actually are.

A full audit of your search and AI presence: which prompts your category owns, who is being cited, where you rank, what’s decaying. The unflattering version, before the fix.
Diagnostic · your domain
delivered · day 3
Authoricy Score
Mid-market SaaS · buyer’s funnel
42/100
SEO presence68solid
AI search visibility31behind
Information gain28low
Citation worthiness39fixable
Decay risk · 6mo12 of 247 pagesflagged
§ 02Strategy

A 30-day roadmap, not a 90-page deck.

Three editorial beats, opened against the gaps the diagnostic found. Each one carries its own keyword queue, so the system always knows what to write next and never writes the same thing twice.
Content plan · your domain
3 beats live
30-day roadmap
Your company · Q3 2026 editorial plan
Week 1Fix indexation. Open 3 beats: pricing, RevOps, buyer psychology.
Week 2First two briefs · usage-based pricing data, attribution myths.
Week 3Refresh 12 decaying pages. Re-index. Schema upgrade.
Week 4Publish piece one. Source 4 expert quotes for piece two.
3 beats · 6 briefs · 12 refreshesdelivered as a working document
§ 03Optimise

We engineer for how AI actually retrieves.

Two methods, combined. Topical clusters build pillar pages with enough semantic depth that engines read authority on a whole beat rather than a single page. Sandbox fan-out tests each piece against the sub-queries an AI would generate from it, before it ships.
Method · topical depth + retrieval test
proprietary
Topical cluster · pricing beat01 · structure
Usage-basedPer-seatHybridTieredFreemiumEnterpriseSaaS Pricing
Sandbox fan-out · retrieval simulation02 · test
“best pricing model for B2B SaaS”
usage vs per-seat retention
tiered SaaS benchmarks
freemium NRR data
hybrid pricing case studies
Predicted retrieval coverage14 / 16
The shift

Rankings show where you were. Citations show where you are.

§ 04Brief

Briefs as architecture, not topic lists.

Every piece starts as a brief: the angle nobody’s taken, the original data we’re sourcing, the experts on call, the decay curve we’re countering. By the time a writer sees it, the thinking is done.
Editorial brief · SaaS Pricing
v2 · approved
Beat · Q3 2026
Usage-based pricing: what the retention data actually says
Topical depthhigh
Newsworthythis week
Info gapretention ≠ growth
Experts2 quoted
Outline
The retention paradox · primary data
Why NRR lags in Q1–Q2
Expert commentary · Kovacs, Lindqvist
Operator playbook · 4 levers
§ 05Produce

Graded against the brief, before anyone sees it.

The draft is scored against the brief it came from: entities covered, questions answered, structure sound. It revises itself until the score stops improving. Anything still short of the bar is held as a draft with the reason attached, and never reaches your site.
Article · published
12 min
SaaS · Pricing
How usage-based pricing actually retains customers
Authoricy Editorial·Fact-checked·3 expert sources

Usage-based pricing is growing 38% faster than per-seat across mid-market SaaS, per a Q3 survey of 340 finance leaders commissioned for this piece.

“Most retention data is survivorship-biased,” says Dr. Anna Kovacs at SSE. “The ones still reporting are the ones that made it work.”

Original researchExpert verifiedFact-checked
The state of search · 2026
58%
of Google searches now end without a click.
Similarweb · Q1 2026
+1,200%
growth in AI-engine referrals to publishers, year over year.
Adobe Analytics · Mar 2026
79%
of B2B buyers consult an AI engine before contacting a vendor.
Gartner · 2026
§ 06Distribute

Straight onto your own site.

Published straight to your own site — WordPress, EmDash, a git repository, or any endpoint you write — with tags and structured data already attached. The link network, when it opens, will earn links on relevance rather than swapping them back and forth.
Distribution log · +48h
auto-tracked
Usage-based pricing studyposted 14 May
LinkedIn · founder + 4 operators42K reachlive
X · thread + data card18K impressionslive
Newsletter · The Operator’s Brief8.4K subslive
Syndication · SaaS Mag, RevOps Today2 placementsin review

It gets better
the longer it runs.

The engagement compounds

Every cycle deepens what we know about your market.

A.The system learnsYOUR COMPANY · YOUR MARKET · YOUR BUYERS01DiagnoseAudit. Score.02Strategy30-day roadmap.03BriefAngles. Experts.ProduceWriters. Editors.04DistributeFirst 48 hours.0506MeasureWhat got cited.

Most content programmes run on guesses. This one runs on a ledger. Every article it publishes, holds or skips is recorded with the reason, and tomorrow’s decision reads that record: which keywords are spent, which beats are behind, what cleared the bar last month and what did not.

By month three, our briefs are sharper than your in-house team’s. By month six, we know what to publish next week before you do. That’s the compounding.

Two ways to work with us

One you can buy today. One we build inside your company.

Autopilot

Thirty articles a month, researched, written and published to your site with nobody in the path. Six checks run before anything goes live, and one of them is that every link has to be a page the system actually opened. It tells you which articles it held and why. WordPress and EmDash.

How it works →
AI Lab

Most companies have a stalled pilot rather than a working system. We build the internal tools, the workflow automation and the closed loops that survive contact with production, and we ship in weeks. Mid-market and enterprise.

What we build →
Pricing

One price for the product. A conversation for the rest.

AutopilotBuy today
EUR129
per month
Thirty articles a month, published to your own site. Connect WordPress or EmDash and the system takes it from there.
30 articles / month, auto-published
Held below the quality bar, never shipped weak
Every run explained: live, held or skipped
Keyword research and briefs included
Non-reciprocal member backlinks
See how it works →
One site. Cancel whenever you like.
AI Lab
Scoped
per engagement
Internal tools, workflow automation and closed-loop systems, built for your stack. Priced against the work rather than a rate card.
Closed-loop audit to start
Internal tools and agent workflows
The integration layer nobody wants to build
Shipped in weeks, not quarters
Your engineers keep the keys
Talk to us →
Mid-market and enterprise.
From the editors

Thinking on AI-cited content.

All articles →
Guide

Enterprise AI SEO: How Large B2B Companies Build AI Search Visibility at Scale in 2026

Enterprise AI SEO delivers 748% ROI when scaled across business units. Complete guide covering multi-platform citation strategy, cross-functional coordination, budget allocation, and the 90-day enterprise implementation framework.

14 min readPRISM 8.5
Guide

How AI Search Engines Select Sources: The Retrieval and Citation Mechanics B2B Marketers Must Understand

AI search engines use a two-stage process: retrieval selects candidates, citation scoring decides what gets referenced. 80% of cited URLs rank outside Google top 100. Here is the mechanism.

13 min readPRISM 8.7
Guide

Search Everywhere Optimization: The B2B Guide to Multi-Platform AI Visibility in 2026

Search Everywhere Optimization captures the 26.3% of buyer research happening outside Google. Complete B2B guide covering AI platforms, Reddit, YouTube, review sites, and the measurement framework for multi-surface visibility.

14 min readPRISM 8.6