The archive grows. Its value doesn't.
Pages compete, decay or disappear. The best answer becomes hard to find.
Content · Search · AI knowledge systems
Most content problems are structure problems.
I'm Joel Falconer. I diagnose the architecture behind large content and knowledge operations — what should exist, how people and machines find it, where it leads and whether the team can sustain it.
If writers, SEO or AI have produced activity without authority, demand or revenue, production is not the constraint. I find the structural one and rank the decisions that resolve it.
Start a diagnostic
Thirty-minute fit call · Australia / worldwide
2.5Bweb and email impressions · SitePoint, 2016–22
17,000+articles under editorial and search operations
Pre-ChatGPTmachine-assisted intake and curation in production
Operating history
SitePointManaging Editor · archive, audience, product
The Next WebFeatures Editor · emerging technology
EnvatoEditorial and marketplace operations
AppStorm + DesignCrowdAcquisition and integration contexts
Live position · 2026
The constraint moved from production to judgment.
AI made production abundant. It did not decide what deserves to exist, which evidence supports it, how it will be retrieved or when it should be removed.
Evidence before assertion
Retrieval before volume
Operators before automation
When to call
The system is producing. It is not compounding.
Four symptoms of the same structural failure.
AI raised output. Authority fell.
Production accelerated without stronger evidence, distinction or a reason to cite you.
Traffic arrives. It leads nowhere.
Search captures demand without creating product value or an owned relationship.
The system works until one person leaves.
Decisions live in people's heads. Standards drift and ownership blurs.
What you get
You leave with decisions the team can run.
A bounded engagement resolves what changes, in what order and who owns it.
01
The decision map.
Keep, consolidate, rebuild, reroute, prove, defer or remove — ranked by leverage and dependency.
02
The retrieval architecture.
The evidence and pathways that let people and AI systems find and trust what matters.
03
The operating specification.
Ownership, workflow, AI boundaries, review gates and measurement — written for the team to run.
Best fit: a live archive or knowledge operation, access to evidence and an owner who can act.
Test the fit
Operating record
Twenty years inside the systems, not outside them.
SitePoint · Managing Editor, 2016–222.5B impressions. 17,000+ articles. Operating scale changed the question.
Across web and email: 2.5 billion impressions and periods above 1.6 million weekly pageviews. The job was deciding what to preserve, connect, consolidate, redirect — and stop carrying.
Taxonomy, search and retrieval9,000 legacy tags became a governed process.
I turned a deferred taxonomy problem into rules, controlled batches and editorial QA — the same governance problem now appearing in semantic retrieval.
Machine-assisted operations · SitePoint, 2020–22Machine learning entered the workflow before the category arrived.
Before ChatGPT, editorial intake and curation connected forms, automation and machine classification. Repetition moved faster; judgment stayed explicit.
Envato, AppStorm + DesignCrowdEditorial properties had to work as transferable systems.
The asset was not content alone, but the audience, operating logic and evidence that could survive the people who built it.
Building now
The method now runs as software.
Current implementation work in provenance, retrieval and semantic governance.
Weave Genesis
A local-first system that turns messy corpora into retrievable claims without losing their sources, with uncertain taxonomy kept under operator control.
Coral system
The system behind this page: one semantic component layer across eight themes and four kinds of surface.
Start with the constraint
Bring me the system that stopped making sense.
Send the problem in a sentence or two. I will tell you whether a thirty-minute diagnostic is the right next step — and what decision it should resolve.