The Meridian Model
A Discipline for Dependable AI Work
The Meridian Model is RBCG's discipline for building reliable software when AI is part of the workflow: engineering foundation, failure recognition, human-assisted method, and the boundary between creative and operational AI work.
AI changes the work. It doesn't remove the need for discipline.
When AI enters your workflow, the failure modes change, but the need for rigor doesn't disappear. The Meridian Model provides the vocabulary, structure, and discipline to build dependably when AI is part of the system.
The Model
Four works, one discipline
The Meridian Model connects dependable software engineering, AI failure recognition, human-assisted development practice, and the boundary between creative and operational AI work.
Foundation
Core Software Engineering
The baseline discipline for building software that is understandable, testable, safe to change, and measurable. This is where all reliable work starts, before AI enters the picture.
Read the publication
Failure Recognition
Naming What Can Fail
A structured vocabulary for failure modes in AI-assisted work: hallucinations, drift, prompt fragility, and more. You can't manage what you can't name.
Read the publication
Method
Human-Assisted AI Development
The practice of working with AI as an assistant: verifying output, building incrementally, maintaining human oversight, and avoiding over-delegation.
Read the publication
Boundary
Creative vs. Operational AI
The critical distinction between AI work that explores possibilities and AI work that must behave reliably in production. Different work, different discipline.
Read the publication
Who It's For
Built for the people who own the outcome
Whether you're leading engineering, building product, or governing AI use across the enterprise.
The Problem
Your teams are building with AI but there's no shared vocabulary for what 'production-ready' means when AI is involved.
Our Answer
The Meridian Model provides the engineering vocabulary and discipline framework to set clear standards for AI-assisted work.
The Problem
You're shipping AI features, but you're not sure which ones need operational rigor and which ones can tolerate creative variability.
Our Answer
The creative/operational split gives you a clear framework for deciding where to invest in reliability and where to embrace exploration.
The Problem
You're building shared AI infrastructure but teams are using it in wildly different ways with different expectations.
Our Answer
The Meridian Model provides shared language for classifying AI work and matching infrastructure to actual requirements.
Creative vs Operational
The work decides the discipline.
Most production systems contain both domains. The mistake is applying one discipline to the whole system.
Creative Domain
AI work that explores, generates, and assists human judgment. The goal is possibility, not repeatability. Output is reviewed, refined, and curated by humans.
Core question
Did this help me think, write, or create better?
Discipline
Iteration, human review, tolerance for variation. Success is measured by the quality of the human-curated result, not the raw AI output.
Operational Domain
AI work that must behave reliably, consistently, and at scale. The goal is dependability. Output goes directly to users or downstream systems.
Core question
Will this behave correctly at 3am with no human watching?
Discipline
Evaluation suites, regression testing, monitoring, rollback capability. Success is measured by reliability, latency, cost, and failure rate.
About Us
The Meridian Model governs everything RBCG delivers.
This isn't a framework we recommend. It's the discipline we apply to every engagement. When RBCG helps you build, run, and improve AI systems, the Meridian Model is the foundation that guides how we classify work, manage risk, and ensure production reliability.
Engineering Foundation
Every system starts with understandable, testable, measurable software, before AI enters the picture.
Failure Vocabulary
We name failure modes explicitly so teams can manage them deliberately.
Domain Awareness
We match discipline to domain, giving creative work creative latitude and operational work operational rigor.
Read the full model
RBCG explains the discipline. Meridian holds the depth.
The dedicated Meridian Model site contains the articles, publications, and PDF request flow for the full body of work.
Meridian Model Home
The dedicated site for the full body of work: articles, publications, and the complete model framework.
Open on meridian-model.ai
Articles
Essays and commentary on applying the Meridian Model to real engineering challenges.
Open on meridian-model.ai
Publications
The formal publications that define the four works of the Meridian Model. Request PDF access.
Open on meridian-model.ai
Get In Touch
Ready to apply the Meridian Model?
Talk to RBCG about how we apply this discipline to enterprise AI engagements.
We typically respond within 1–2 business days.