@ben_burtenshaw: this is my working categorization for model shops in 2026, updated for open weight and / RSI. concepts like 'pre-produc…
Summary
Ben Burtenshaw outlines a non-sequential categorization for AI model shops in 2026, covering phases like pre-blog, pre-lab, pre-model, pre-rsi, and pre-cloud, emphasizing the need for updated thinking beyond traditional 'pre-product' concepts.
View Cached Full Text
Cached at: 07/12/26, 06:59 PM
this is my working categorization for model shops in 2026, updated for open weight and / RSI.
concepts like ‘pre-product’ don’t make sense, so we need to think more about blogs, models, labs, RSI, and cloud.
- pre-blog post: they’ve announced but have no blog posts doing non-novel research, just with visuals. (note this is also my personal favorite phase to work with)
- pre-lab: they’ve got sick blog posts, but you’re not sure if they’re a lab or just a blog shop. (also, love consuming this content)
- pre-model: they’re doing amazing stuff. it sounds impressive. but you don’t know if it will become a model, a product, or more blog posts. (my least favorite phase)
- pre-rsi: they’re shipping, and it works, but you don’t know if that’s the output of their thing or the thing itself.
- pre-cloud: they have models and blog posts. win! they do rsi. but you don’t know whether they’ll become a cloud provider. (struggle to relate to this phase. cloud is boring…)
note, these phases are not sequential. labs can be pre-blog post. rsi shops can be pre-model.
Similar Articles
@mattpocockuk: Feels like categorising models got harder recently I used to put models in the bucket of Opus-like, Sonnet-like, or Hai…
Matt Pocock observes that categorizing AI models has become harder with new model names like Fable and shifting performance tiers, and asks the community how they evaluate models.
@DeRonin_: As an AI engineer in 2026, learn this: > systematic output reading. pattern recognition across 1,000 model responses is…
A seasoned AI engineer shares key skills for 2026, including systematic output reading, context engineering, tool description discipline, eval design, model routing, prompt versioning, confidence scoring, streaming architecture, fallback chains, latency budgets, failure cataloguing, agent-vs-workflow decisions, and failure post-mortems as portfolio content.
@mattpocockuk: Here are my 7 phases of AI-powered development. I've been thinking that the pre-PRD phase needs more structure. You nee…
Matt Pocock shares his seven phases of AI-powered development, emphasizing the need for more structure in the pre-PRD phase.
@rhythmrg: https://x.com/rhythmrg/status/2066561780495896785
The article argues that enterprises should post-train their own custom AI models for mission-critical, high-volume use cases to achieve differentiation, cost savings, and control over tradeoffs, rather than relying solely on general frontier models.
@vipulved: https://x.com/vipulved/status/2071404852908081211
An essay arguing that the AI ecosystem is undergoing modularization similar to the PC revolution, with standardized interfaces like transformers, inference APIs, and agentic harnesses enabling specialization and rapid innovation, and that open-weights models are a direct economic consequence.