Ryan Bishop Consulting · A learning journey

From silicon to inference.

Every answer from a language model begins as refined sand — and crosses half the world before it reaches you. Follow the twenty-six steps from raw silicon to a finished thought.

26 stages · ~10 minutes · sound optional
Our abstract, “Current GenAI Policy In HTA: The Missing Mandate And Implications For Industry,” has been accepted for poster presentation at ISPOR Europe 2026 in Vienna.
Now entering
Key characteristics
In practice
Drag to explore→
Journey layer
Geopolitical
Hardware
Software
Model Creation
Inference
The journey, end to end

That's the journey behind a single answer.

"How do language models actually work?"
SiliconInference
A finished thought — drawn from refined sand, across three continents and twenty-six steps.

Twenty-six stages · three continents · one sentence. Every layer earns its place before a model can answer you.

And beyond
🤖💬
Agentic AI
Once a model can answer, the next step is letting it act — planning, using tools, and pursuing goals across many such answers. That is where the journey leads next.
Ryan Bishop Consulting
Ryan Bishop Consulting
ryanbishop.co.uk
Explore all 125 cards →