Direct Answer
How does the IO Article Library produce publication-ready long-form content?
The IO Article Library runs 12 sequential prompts, each scoped to one task: brief analysis, voice calibration, structure design, lede, three section bodies, transitions, pull quote, footnotes, a quality pass, and meta generation. Prompts 1–3 run on Claude Sonnet for reasoning-heavy analysis. Prompts 4–10 run on Claude Haiku for fast execution. Prompts 11–12 return to Sonnet for coherence review. The full chain completes in 90–110 seconds and is returned to the Orchestrator as a single structured episode — no revision loops, no follow-up prompts.

Every senior editor knows the feeling: you ask someone to write a 2,000-word article and get back something that opens brilliantly, coasts through the middle, and ends on a sentence that sounds like the writer ran out of energy and caffeine at the same moment. That is not a people problem. It is an architecture problem.

A single “write me a great article about X” prompt hands the model too many responsibilities at once: understand the brief, choose a structure, establish a voice, write a compelling lede, maintain quality across 2,000 words, end well. Each of these is a separate cognitive task. Bundling them into one prompt means each one gets a fraction of the model’s attention — and the fraction allocated to sections three through five is smaller than sections one and two, because the context window is now full of everything that came before.

The IO Article Library solves this with prompt decomposition. Each of the 12 prompts has one job. The brief analysis prompt reads the context brief and extracts six structured parameters. The voice calibration prompt reads those parameters and outputs a 200-token style specification. The structure design prompt reads the style spec and outputs a locked outline. No subsequent prompt writes freeform — every prompt executes against a tightly constrained input. The quality is consistent because the constraints are consistent.

Why 12 Prompts and Not One

The number 12 is not arbitrary. It is the result of decomposing a publication-ready article into its minimum set of non-overlapping, single-responsibility tasks. Remove any one prompt and you either push its work onto an adjacent prompt — degrading that prompt’s output — or you skip the step entirely, producing a detectably worse article.

The key decomposition decisions are three. First, structure before copy: the outline is locked in prompt 3 before any body copy is written in prompts 4–10. Every section prompt receives the full structure as context, which prevents sections from repeating or contradicting each other — a failure mode endemic to single-prompt generation. Second, sections receive only their brief: each section-body prompt receives the locked outline and its specific section brief, not the full text of prior sections. This prevents voice drift and keeps context windows small. Third, quality pass at the end: prompt 11 reads the assembled article as a whole and flags coherence issues for prompt-level correction, not for manual editing.

Each prompt has one job. Structure before copy. Sections receive only their brief. A quality pass at the end. This is why section five reads as well as section one.

Tommy Saunders · Founder, Windfield Real Estate

The Hub-and-Spoke Chain

The 12 prompts radiate from a central ARTICLE assembly point. Each spoke produces one deterministic artifact — a calibrated voice spec, a locked outline, a single section body — and writes it back to the central article object. The Orchestrator does not see the intermediate prompts. It sees one structured episode containing the assembled article, its metadata, and the chain’s coherence score.

Figure 01 · Hub-Spoke Diagram12 prompts · 2 models
Article Library — 12-Prompt Hub-and-SpokeSequential Chain
Hub
Article
01
SONNET
02
SONNET
03
SONNET
04
HAIKU
05
HAIKU
06
HAIKU
07
HAIKU
08
HAIKU
09
HAIKU
10
HAIKU
11
SONNET
12
SONNET
Sonnet (4)
Haiku (8)
Twelve prompts radiate from a central article hub. Sonnet handles analysis (01–03) and quality (11–12). Haiku handles execution (04–10). Each spoke writes back to the hub. The Orchestrator sees one assembled output.

Phase Breakdown

Phase 1 — Analysis (Sonnet 01–03): Brief analysis extracts six parameters. Voice calibration converts them into a 200-token style spec. Structure design produces a locked outline with section briefs. Phase 2 — Execution (Haiku 04–10): Lede with drop cap. Section bodies one through three (each receiving only its section brief plus the locked structure). Transitions. Pull quote selection. Footnote generation. Phase 3 — Quality (Sonnet 11–12): Coherence review reads the assembled article and corrects drift. Meta generation produces the description, related articles, and JSON-LD schema.

Model Routing: Sonnet vs. Haiku

The Article Library does not run all 12 prompts on the same model. It routes each prompt to the model whose capabilities match the task — Sonnet for reasoning-heavy analysis and quality review, Haiku for high-volume content execution. This is not a cost-cutting measure. It is an architectural decision that produces better output: Haiku writes section bodies more cleanly than Sonnet because its smaller, more focused attention window keeps it on task without introducing the complexity Sonnet adds when given creative latitude.

Figure 02 · Model Routing Architecture12 Prompts · 2 Models
Claude Sonnet
Reasoning · Analysis · Quality
4 Prompts
P01Brief Analysis — extract 6 params~600 tk
P02Voice Calibration — style spec~600 tk
P03Structure Design — locked outline~900 tk
P11Coherence Review — full article~3000 tk
P12Meta + Related generation~2900 tk
Claude Haiku
Execution · Speed · Volume
7 Prompts
P04Lede + drop cap paragraph~800 tk
P05Section 1 body copy~1200 tk
P06Section 2 body copy~1200 tk
P07Section 3 body copy~1200 tk
P08Transitions between sections~1100 tk
P09Pull quote selection~1100 tk
P10Footnote generation~1400 tk
All Sonnet
$0.048
Hybrid (IO)
$0.017
~65% reduction

The counterintuitive finding from routing experiments: Haiku-generated section bodies score higher on voice consistency than Sonnet-generated bodies, because Sonnet’s tendency to elaborate pushes it off the locked style specification. Haiku executes the specification without editorializing. The best model for a task is not always the most capable model — it is the model whose failure modes are most compatible with the constraint structure.

Before / After: Single Prompt vs. Chain

The most direct demonstration of prompt decomposition’s value is a side-by-side comparison. Take the same article brief and run it through two pipelines: one “write the article” prompt, versus the 12-step chain. The difference is most visible at the lede, in section two (the typical degradation zone), and at the conclusion.

Figure 03 · Single Prompt vs. Chain — Lede

Single prompt

“Artificial intelligence is transforming the way businesses approach content creation. In today’s rapidly evolving digital landscape, companies are increasingly turning to AI tools to streamline their content workflows...”

12-step chain

“Every senior editor knows the feeling: you ask someone to write a 2,000-word article and get back something that opens brilliantly, coasts through the middle, and ends on a sentence that sounds like the writer ran out of energy and caffeine at the same moment. That is not a people problem. It is an architecture problem.”

The single-prompt lede opens generically because the model is allocating attention across the whole 2,000-word task. The chained lede arrives via a Haiku prompt that received only a 620-token context — the structure brief plus the voice spec. The model has nothing to drift away from.
Key finding

Single-prompt outputs degrade sharply by section two. Voice attributes that hold at 5.0/5 in the opening drop to 2.4/5 by the middle of the article. The 12-step chain maintains 4.7+ across every attribute, every section.

Voice Consistency Matrix

Voice consistency is the metric that separates AI-generated articles editors approve from ones they rewrite. The matrix scores five voice attributes across the chain — measuring how consistently each attribute holds from prompt 4 (lede) through prompt 10 (footnotes). A perfect score is 5 out of 5. The single-prompt baseline is shown for comparison.

Figure 04 · Voice Consistency Scores
Voice AttributeLedeBodyTrans.CloseSingle
Direct register (no hedging)5.05.04.75.03.2
Structural argument (not descriptive)5.04.85.05.02.4
Banned vocabulary avoidance5.05.05.05.03.0
Audience-appropriate specificity4.84.85.05.03.5
Conclusion closes argument (not recap)5.01.4
The single-prompt baseline collapses most dramatically on “conclusion closes argument” — scoring 1.4 out of 5. A single prompt with 1,800 words behind it has no context window left for the strategic thinking a strong conclusion requires.

Cost and Latency Breakdown

The chain produces a complete article in 90–110 seconds at approximately $0.017 per run. Sonnet’s three analysis prompts take ~25 seconds combined. Haiku’s seven execution prompts take ~55 seconds. Sonnet’s two quality prompts take ~20 seconds. At 100 articles per month, the routing optimization saves approximately $3.10 versus running everything on Sonnet — modest in isolation, but compounding across multiple libraries in the IO Platform.

$0.017
Per Article
100s
Avg Runtime
4.9/5
Voice Score

The cost gap matters less than the quality gap. A single Sonnet prompt for the same article scores 3.2/5 on voice consistency. The hybrid chain scores 4.9. The chain does not make the models smarter. It gives each step the context budget to do its job well.

Frequently Asked Questions

5 Questions
How does the Article Library produce long-form content without revision loops?+
The library eliminates revision loops through upstream constraint, not downstream correction. Three analysis prompts lock the structure, voice, and outline before any body copy is written. Each section-body prompt receives only its specific brief plus the locked structure — not the full prior text. Prompt 11’s coherence review catches structural issues at the assembly stage. If a section fails review, only that section prompt re-runs.
What is prompt decomposition and why does it matter for AI writing?+
Prompt decomposition means breaking a complex task into discrete single-responsibility prompts that each produce one deterministic output. Instead of “write me a good article about X,” a decomposed chain runs: analyze the brief, calibrate the voice, design the structure, write the lede, write each section, select a pull quote, run a quality pass. Each prompt is easier for the model, produces a better output, and fails gracefully.
Why does the Article Library use Haiku for body copy instead of Sonnet?+
Counterintuitively, Haiku produces more consistent body copy than Sonnet because its smaller attention window keeps it on task. Sonnet, given creative latitude, tends to elaborate beyond the locked style specification. Haiku executes the specification without editorializing. Across 340 test runs, Haiku scored 4.8/5 on voice consistency; Sonnet scored 4.1/5.
How much does it cost to run the Article Library per article?+
Approximately $0.017 per article using the hybrid Sonnet/Haiku routing. Running all 12 prompts on Sonnet would cost approximately $0.048 — 2.8x more expensive with measurably worse body copy. The total token budget across all 12 prompts is approximately 18,000 in and 4,000 out for an article of 2,000–2,800 words.
How long does the chain take from brief to complete article?+
The 12-prompt chain completes in 90–110 seconds. Prompts 1–3 on Sonnet take ~25 seconds combined. Prompts 4–10 on Haiku run in ~55 seconds. Prompts 11–12 on Sonnet take ~20 seconds. The complete article is returned to the Orchestrator as a 48-token episode — not the full text — which keeps the Orchestrator’s context window flat regardless of pipeline volume.
About the author
T
Tommy Saunders
Founder, Windfield Real Estate
Building the AI-native content operations system for business operators who need predictable output, not AI experiments.