Zero to MVP Weekly

← What this newsletter is

📬 Every Saturday · ~5 min read

Issues

One signal, one tool, one MVP idea — every Saturday. Every issue is free to read; subscribe to get the next one in your inbox.

Issue #9 · August 8, 2026 · ~5 min read →

The worm that earned its green check

Signal: an npm worm with legitimately earned provenance planted hooks in .claude and .vscode config
Tool: LFM2.5-2.6B — a real agent model in 2.5 GB of RAM, no GPU, no cloud
MVP idea: a "who can run code on my machine?" repo auditor

Issue #8 · August 1, 2026 · ~5 min read →

MCP just became boring infrastructure

Signal: MCP went stateless — exposing your product to agents now costs about what a REST API costs
Tool: GPT-5.6 Luna at 80% off — a hands-on read on real coding tasks
MVP idea: a "confirm before it bites" MCP server template

Issue #7 · July 25, 2026 · ~5 min read →

Frontier performance, half the price

Signal: four vendors in five days, all priced on cost per finished task
Tool: Claude Opus 5 — a hands-on look on real code
MVP idea: cost-per-task receipts for your agent runs

Issue #6 · July 18, 2026 · ~5 min read →

Open weights went frontier this week

Signal: Kimi K3's 2.8T vs Murati's Inkling — "open weights" stopped meaning one thing
Tool: Kimi K3 via the API — a first hands-on look
MVP idea: a correction-data logger for coding sessions

Issue #5 · July 11, 2026 · ~5 min read →

Four coding models in one week

Signal: GPT-5.6 in three tiers, Grok 4.5 at $2 — list price stopped being comparable
Tool: Grok 4.5 inside Cursor — a hands-on verdict
MVP idea: a cost-per-task meter for coding agents

Issue #4 · July 4, 2026 · ~5 min read →

The week your model became a policy risk

Signal: Fable 5 returns from a 19-day export-control blackout — and the market hedges
Tool: Claude Sonnet 5 — near-Opus agentic work at a third of the price
MVP idea: a model continuity kit — "what's your fallback?"

Issue #3 · June 27, 2026 · ~5 min read →

The week the industry stopped paying for wasted tokens

Signal: GPT-5.6 lands, but the real shift is the industry unwinding "tokenmaxxing"
Tool: Google AI Studio Starter Tier — deploy an app with no credit card
MVP idea: a token-spend co-pilot for small dev teams

Issue #2 · June 20, 2026 · ~5 min read →

Open weights just hit the cheap frontier for coding

Signal: an open model (GLM-5.2) beats GPT-5.5 on coding at ~1/6 the cost
Tool: GLM-5.2 vs Kimi K2.7 Code — a hands-on verdict
MVP idea: codemod-as-a-service built on cheap open models

Issue #1 · June 14, 2026 · ~5 min read →

Apple's on-device AI changes the math for small apps

Signal: Apple's on-device AI is now free for apps under 2M downloads
Tool: MiniMax M3 — an honest look at the open frontier model
MVP idea: a pay-once receipt scanner built on free on-device AI