DEV Community: Mohan Iyer

The Five-Model Consensus Architecture Behind ConsensusPress WordPress Plugin

Mohan Iyer — Sun, 24 May 2026 06:09:51 +0000

Most AI writing tools are a single API call dressed up with a nice UI.
One model. One output. Publish and hope.
ConsensusPress AI Publisher, a WordPress plugin that validates content across five AI models simultaneously, is architecturally different. This post is the technical explanation of why — and what the trade-offs are.
Why does running five models produce a stronger signal than one?
Five independent models produce a stronger signal than one because their disagreements are as informative as their agreements — and no single model can surface its own blind spots.
GPT, Claude, Gemini, Llama, and Mistral were each trained on different datasets, with different fine-tuning objectives, different safety frameworks, and different confidence calibration. What one model treats as settled, another may flag. What one model considers well-sourced, another may question.
When you run all five against the same query in parallel, two things happen. First, the overlap between their outputs identifies high-confidence claims — the things five independent systems trained differently all agree on. Second, the divergence identifies contested zones — the claims worth surfacing to a human editor before they publish.
Independent generation before synthesis is the critical design choice. Each model generates its draft without seeing the others. This prevents anchoring bias — the tendency of models shown a prior output to converge on it regardless of its quality.
How does the consensus engine score and route outputs?
The consensus engine evaluates five independent drafts using semantic similarity scoring, claim extraction and overlap analysis, majority voting, and weighted confidence scoring — then routes the result through a tiered decision matrix.
The routing logic works as follows:
Five of five models in agreement — the post clears that dimension and routes to the editor as a validated draft.
Four of five models in agreement — an advisory flag is raised with the specific concern noted. The draft proceeds but the flag is visible to the editor.
Three of five models in agreement — a soft block is triggered. The post cannot publish without editor review and sign-off.
Two or fewer models in agreement — a hard block. The post cannot publish without a deliberate override and a documented reason.
The result is a consensus confidence matrix rather than a binary pass/fail. Divergence is not suppressed — it is surfaced. A claim that four of five models flag is a high-confidence editorial risk signal. A claim that only one model flags is worth investigating but not auto-blocking.
This distinction matters in production. Blunt gates that block on any disagreement generate false positives and editor fatigue. Tiered routing preserves editorial judgment while ensuring nothing significant slips through silently.
How does it integrate with WordPress and what are the real trade-offs?
ConsensusPress sits in the WordPress editor sidebar, returning results in under sixty seconds via parallel async API calls to all five models — with only the synthesis step requiring sequential processing.
The WordPress integration delivers three things to the editor: the consensus draft, divergence flags with the specific claims that triggered them, and a confidence score per validation dimension. Every published post carries a permanent audit trail of its consensus scores and any overrides, with override reasons documented.
The trade-offs are real and worth stating honestly.
API cost is the most obvious. Five simultaneous model calls cost more than one. Mitigation: caching frequent queries and using the consensus layer selectively on high-stakes content rather than every draft.
False consensus is the structural risk. Models trained on substantially overlapping corpora will have correlated blind spots. Five models agreeing does not guarantee correctness — it guarantees that five systems with different architectures found no disagreement on this claim. That is a meaningful quality signal, not an infallible one. Retrieval-augmented generation grounding outputs in authoritative sources is the mitigation — not a complete solution.
Latency is manageable. Parallel execution means the generative stage runs in the time of the slowest single model call, not the sum of five. Under sixty seconds is achievable on current infrastructure for standard post lengths.
Vendor lock-in risk is real over a multi-year horizon. The abstraction layer between the consensus engine and individual model APIs must be maintained as model versions update silently and prompt sensitivity to structured formats changes between releases.
What does this mean for developers building on top of it?
The architecture is modular by design. Models can be swapped or weighted differently by content type — legal content may weight Claude higher for safety alignment, creative content may weight GPT higher for generative range. The consensus threshold itself is configurable.
The divergence flag layer is where the most interesting extension work lives. The current implementation surfaces flags to human editors. The natural extension is automated routing by flag type — factual accuracy flags to a fact-checking queue, bias flags to an editorial review queue, tone flags to a copy editor queue.
The audit trail — every post carrying its consensus scores and override history — is the compliance play for regulated industries. Healthcare, finance, and legal content publishers can demonstrate documented AI validation as part of their editorial due diligence.
ConsensusPress is live on WordPress.org. Free tier available — three posts per month, no credit card required.
The re-anchor template used to build this across twenty-plus AI sessions is open source: github.com/Mohan-Iyer/re-anchor-template

This article was generated using ConsensusPress AI Publisher — five-model consensus, hallucination-filtered. Champion: Mistral, 100/100. Agreement level: 50% across five models.

Title: The 44% Problem: What Google Search Console Taught Me About AI Consensus Content Canonical URL: https://seekrates-ai.com/google-indexing-half-your-content-the-44-problem/ Tags: seo, wordpress, webdev, ai Note: Cross-post from seekrates-ai.com. Set

Mohan Iyer — Thu, 05 Mar 2026 23:37:15 +0000

Google Indexing Issues Explained The 44% Problem

Discover how Google indexes content and what the 44% problem means for your website's visibility. Explore solutions with Seekrates AI.

seekrates-ai.com

When 5 AI models split 50/50 on a question, that's not a failure of the platform — it's the most honest answer possible. Our 201-post consensus index at seekrates-ai.com shows that sub-70% consensus questions fall into two categories: genuinely contested (

Mohan Iyer — Thu, 05 Mar 2026 05:32:45 +0000

GEO vs Traditional SEO: What Changes When AI Answers the Question

Mohan Iyer — Sun, 15 Feb 2026 02:10:30 +0000

This article was originally published on Seekrates AI. For the methodology behind how we create AI-validated content, see The Re-Anchor Manager.

How AI Language Models Decide Which Sources to Cite

Mohan Iyer — Sun, 15 Feb 2026 02:08:34 +0000

This article was originally published on Seekrates AI. For the methodology behind how we create AI-validated content, see The Re-Anchor Manager.

llms.txt: The File Your Website Probably Doesn't Have

Mohan Iyer — Tue, 03 Feb 2026 00:26:35 +0000

You have a robots.txt. You probably have a sitemap.xml. But do you have an llms.txt?
What is llms.txt?
The llms.txt specification provides a machine-readable summary of your website for large language models. It's not about blocking crawlers — it's about helping AI understand your content.
Basic Structure
markdown# Your Site Name

One-line description of what you do

Main Resources

Page Title: Brief description
Another Page: Brief description

Topics You Cover

Topic 1
Topic 2
Topic 3 Real Example Here's mine: markdown# Seekrates AI

AI consensus platform that queries multiple AI models and synthesises where they agree

Main Resources

About: Company background and methodology
Blog: AI-era SEO and content strategy
Playbook: AI-Era SEO implementation guide

Topics We Cover

Generative Engine Optimisation (GEO)
Large Language Model Optimisation (LLMO)
Answer Engine Optimisation (AEO)
Multi-model AI consensus methodology Implementation

Create llms.txt in your root directory (same level as robots.txt)
Add your site description and key pages
Keep it under 500 lines
Optional: Create llms-full.txt for detailed documentation

Why Bother?
When someone asks Claude or ChatGPT about your domain, the AI draws from training data and accessible web content. A clear llms.txt is a structured signal about what you do.
Not magic. But a signal most competitors don't have.
Took me 15 minutes to create. Highest signal-to-effort ratio I've found for AI-era SEO.

Full implementation guide with schema markup and Rank Math settings: AI-Era SEO Playbook

GEO, LLMO, AEO: The AI-Era SEO Acronyms You Need to Know

Mohan Iyer — Sat, 31 Jan 2026 06:09:05 +0000

Traditional SEO optimizes for Google crawlers. AI-Era SEO optimizes for machine comprehension. Here’s the difference.

Traditional SEO is dying. Not slowly — now.

40% of B2B buyers now research primarily through AI assistants. Your website might rank #3 on Google, but it’s completely invisible to ChatGPT, Claude, and Perplexity.
Over the past month, I ran an experiment: publishing 139 blog posts in 48 hours, all scoring 80+ on Rank Math’s SEO analyzer. The methodology wasn’t asking one AI to write content. It was asking 5 different AI models the same questions and finding where they agreed.

What I learned: AI-Era SEO rests on three pillars. If these acronyms mean nothing to you, your SEO strategy is already outdated.

GEO: Generative Engine Optimization
GEO is about getting AI search engines to cite your content.

Not just rank you. Cite you.

When someone asks ChatGPT or Perplexity a question, does it quote your site? That’s the new SEO game.

Traditional SEO vs GEO:

TraditionalGEOOptimize for Google crawlersOptimize for AI understandingKeywords in meta tagsSemantic clarity in contentBuild backlinksBuild citation-worthiness

GEO Best Practices for 2026:

FAQ schema on every page
Clear, factual statements (not fluff)
Cite sources AI can verify
Structure for comprehension, not crawling
Answer questions directly
The shift is fundamental. Google rewards pages that match search intent. AI assistants reward content they can trust and cite.

LLMO: Large Language Model Optimization
If GEO is about AI search engines, LLMO is about AI assistants themselves — ChatGPT, Claude, Gemini when they’re answering questions directly.

The goal: Make your content trustworthy enough that LLMs cite it in their responses.

This isn’t about gaming systems. It’s about being genuinely citation-worthy:

Factual accuracy — LLMs cross-reference multiple sources
Clear attribution — They want to cite their sources
Semantic structure — They need to understand, not just parse
Verifiable claims — They’re getting better at fact-checking
The interesting part: What makes content good for LLM citation also makes it good for human readers. Win-win optimization.

The llms.txt file:

You have robots.txt for Google. You need llms.txt for AI assistants.

It’s a simple text file at your domain root that tells AI crawlers:

What to index
How to cite you
What your site is about
Boundaries for AI use of your content
Ten minutes to implement. Significant visibility gains.

AEO: Answer Engine Optimization
AEO predates the current AI wave — it’s about winning featured snippets and now AI Overviews in Google search results.

When Google shows a direct answer at the top of search results, that’s AEO territory. Now, with AI Overviews rolling out, AEO matters even more.

AEO focuses on:

Direct answers to specific questions
Structured data that search engines can parse
Content formatted for extraction (lists, tables, clear headers)
Position zero optimization
AEO is often the quickest win because it builds on traditional SEO skills while preparing you for AI-Era requirements.

How They Connect
These three aren’t separate strategies. They’re layers:

PillarTargetSchema FocusGEOAI search engines (Perplexity)FAQ schemaLLMOAI assistants (ChatGPT, Claude)Article schemaAEOFeatured snippets + AI OverviewsHowTo schema

Schema markup is the bridge. One investment, three returns.

Prioritization (what 5 AIs agree on):

Start with AEO — quickest wins
Build LLMO — long-term authority
Layer GEO — emerging opportunity
All three overlap. Start anywhere. The key is starting.

The Methodology Behind This
Why should you trust this breakdown?

I built a platform that queries OpenAI, Claude, Gemini, Mistral, and Cohere simultaneously — then synthesizes where they agree and disagree.

Single AI responses hallucinate. But when 5 independent models reach the same conclusion, you’re closer to truth.

We call it “Pentangulate” — consensus from 5 perspectives.

Not AI-generated content. AI-arbitrated truth.

The 139 posts I mentioned? Each one validated by this methodology. Each claim checked against multiple AI perspectives before publishing.

What This Means for You
The question isn’t whether to adapt your SEO strategy. It’s how fast.

2020: “How do I rank on Google?” 2026: “How do I get cited by ChatGPT?”

Different question. Different strategy. Different winners.

Most SEO advice is still fighting the old war. GEO, LLMO, and AEO are the new battlefield.

The good news: The fundamentals still matter. Keywords, backlinks, page speed — they’re still the foundation. AI-Era SEO layers on top. You don’t abandon what works. You evolve it.

Start with schema markup. Add an llms.txt file. Structure your content for comprehension, not just crawling.

The shift is happening whether you’re ready or not.

Mohan Iyer is the founder of Seekrates AI, a platform that synthesizes consensus from multiple AI models. Learn more at seekrates-ai.com.
Originally published on Medium