Letter "Dear friends" from Andrew Ng in *The Batch* (DeepLearning.AI, issue 359) on **loop engineering** applied to **0-to-1** product development. Ng shares his **3 key loops** — agentic coding loop (~minutes), developer feedback loop (~hours), external feedback loop (~days) — nested by increasing time scale, connecting *coding agent → product spec/evals → developer vision → external feedback*. Central thesis: humans retain a **context advantage** (rather than a "taste") that makes human-in-the-loop indispensable; engineers take on a partial product management role. Domain: coding agents, product engineering, agentic methodology.
Long-form essay by **Shubham Saboo** (X/Twitter) advancing a thesis on the Product Manager role in the age of agents: the next key skill is **not prompt engineering** but **Loop Engineering** — designing a *system that improves with every run* rather than writing the perfect prompt every time. A **loop** is a repeated cycle: change what shapes the agent's behavior → run it → evaluate the output → keep the change if quality rises, revert otherwise → **compound the learning** so the next version starts ahead. For a PM, the entry point is not code but the **durable artifacts** that encode their judgment: PRD-review skill, customer-call *summarizer*, evaluation rubric, launch checklist, research workflow, `CLAUDE.md`, prompt template, prioritization framework. Because they are reused, these artifacts **compound in both directions** — and **drift** silently (a CLAUDE.md that keeps growing, a checklist that gets ignored…): the model has not regressed, the artifacts have drifted unwatched. A loop has **5 parts**: trigger, action, **proof**, memory, **stop condition** (the most critical). **Evals** become PM work (testing the artifact against known examples: 3 good / 3 bad PRDs, 5 understood calls, 2 past launches). **Memory** lives on **GitHub** (the repo becomes "product memory": commits, diffs, eval results, decision log, rollback). Recommended first loop: a **weekly product signal loop** (every Friday). Taste remains central — but it now needs **proof**. Cites Boris (creator of Claude Code): "he no longer writes prompts, he writes loops."
Op-ed by **Olivier Rafal** (Consulting Director Strategy at **WeNvision**) published on **February 23, 2024** on **CIO-Online** (*Tribune* section), which puts forward a thesis that was still counterintuitive at the time: **generative AI is more a matter of technological product than of an AI/data science project**. **Argument 1 — data science is not the core of the issue**: building a *foundation model* from scratch requires *« several months, millions of euros, and access to enormous quantities of data »* — reserved for players with specific, monetizable datasets (e.g. **Bloomberg** and its **BloombergGPT** for finance). For nearly all companies, the right reflex is therefore not to hire data scientists. **Argument 2 — a skills mismatch**: what's mainly needed are **development and integration engineers** (back/front), **strong cloud skills**, and **DevOps**. Client quote: *« You don't necessarily need to be a data scientist, but you do need to understand the basic concepts, have back-office development skills, and strong cloud skills. »* **Argument 3 — platform architecture (orchestrators + API)**: building an enterprise **plateforme d'IA générative** via orchestrators and API makes it *« easy to work with the best LLMs on the market and switch between them as they each evolve, without reworking the applications »* (anti vendor lock-in). **Argument 4 — from project to product**: *« The platform […] must itself be considered a product »*; instead of a one-off investment, plan for a **monthly funding stream** (continuous iterations, ongoing innovation). **Argument 5 — governance & shadow AI**: the unprecedented democratization of GenAI generates *« as much shadow AI as strong expectations toward the DSI »* → governance to capture business needs, **prioritize products by value**, and oversee proper operation. **Paradigm shift** announced: *« we are moving from classic algorithmic programming to agents Langchain that handle part of the decisions »*. **Relevance for the watch**: a **founding text (2 years ahead)** of WeNvision doctrine (product > project, platform/API, flow-based funding, governance, shadow AI) that the fiches [[wenvision-ai-agents-enterprise-deployment-2025-10-01]], [[habert-ia-agentique-production-2025-10-29]] and [[rafal-wenvision-tokenomics-foundation-finops-ia-2026-06-04]] will extend (FinOps/token, flow-based funding → financial governance). It also prefigures the *harness/platform around the model* (Dropbox/Okumura: *systems around the model*) and **model independence** through an orchestration layer.
#generative AI#technological product#product vs project
**Olivier Rafal** · *Consulting Director Strategy* chez **WeNvision** (cabinet de conseil FR). Tribune publiée dans la rubrique *Tribune* de **CIO-Online**. Auteur déjà présent dans la veille (cf. fiches WeNvision/Atlas/Tokenomics). Publié le **23 février 2024**.