diff --git a/1-planning/stop-switching-ai-image-models.md b/1-planning/stop-switching-ai-image-models.md
new file mode 100644
index 0000000..ba68e7a
--- /dev/null
+++ b/1-planning/stop-switching-ai-image-models.md
@@ -0,0 +1,193 @@
+---
+slug: stop-switching-ai-image-models
+title: "Stop Switching AI Image Models. Pick One and Master It."
+author: henry
+status: planning
+created: 2024-12-29
+updated: 2024-12-29
+content_type: opinion-piece
+primary_keyword: "best ai image generator"
+secondary_keywords: ["flux vs sdxl", "ai image model comparison", "best ai for realistic images"]
+---
+
+# Idea
+
+Henry shares his experience of wasting time hopping between AI image models — Midjourney, DALL-E, Stable Diffusion, Flux, etc. Every new model meant relearning prompts, different strengths, broken workflows. Now he knows: pick one good model and master it.
+
+**Core message:** The "best" model is the one you actually learn to use well. Model-hopping is a productivity trap.
+
+**Why Henry:** 
+- Experienced dev perspective (12 years)
+- Has actually used these tools in production workflows
+- Can speak from real frustration, not theory
+- Establishes him as pragmatic voice in AI tooling space
+
+**Research source:** `/research/trends/top-ai-models-henry-article-2025-12-28.md`
+
+---
+
+# Brief
+
+## Strategic Context
+
+**Why this topic:** 
+Developers are overwhelmed by AI image model choices. Every month there's a "new best" model. Most comparison content is listicles that don't help with the actual decision. Henry offers a contrarian, experience-based take: stop comparing, start mastering.
+
+**Why now:** 
+- Flux 2.0 just released (Nov 2024)
+- Imagen 4 launched (May 2025)
+- Seedream 4.0 topped leaderboards
+- Model fatigue is real — perfect timing for "enough already" message
+
+**Banatie angle:** 
+None explicit in Phase 1. Article establishes Henry's expertise in AI image tooling. Sets up future content about workflow integration (where Banatie fits naturally).
+
+**Henry positioning:**
+This article positions Henry as:
+- Voice of experience ("I've been through this")
+- Pragmatic engineer ("here's what actually matters")
+- Counter to hype cycle ("ignore the leaderboards")
+
+## Target Reader
+
+**Who:** 
+Developer (2-8 years experience) who uses AI image generation for projects — landing pages, prototypes, content. Has tried 2-3 different tools, feels behind on the latest models.
+
+**Their problem:** 
+Constantly seeing "X is now the best AI model" posts. Wondering if they should switch. Worried they're missing out. Spending more time evaluating tools than using them.
+
+**Desired outcome:** 
+Permission to stop chasing. Clear framework for choosing. Confidence in their current choice (or clear reason to switch once, then stay).
+
+**Search intent:** 
+Commercial/Informational hybrid — they're comparing, but also looking for guidance on HOW to choose.
+
+## Content Strategy
+
+**Primary keyword:** "best ai image generator" 
+- Volume: 33,100/mo
+- KD: 31 (achievable)
+- Intent: Commercial — people comparing options
+- Our angle: Subvert the expectation. Not "here's the best" but "here's why that question is wrong"
+
+**Secondary keywords:**
+- "flux vs sdxl" — comparison searchers
+- "ai image model comparison" — direct match
+- "best ai for realistic images" — specific use case
+
+**Competing content:** 
+Mostly listicles: "Top 10 AI Image Generators 2025". Feature comparisons. No one is saying "stop comparing."
+
+**Our differentiation:** 
+- Contrarian angle: "The search for 'best' is the problem"
+- Experience-based: Real workflow friction, not feature lists
+- Decision framework: Not "best overall" but "best for YOUR workflow"
+- Actionable: Ends with clear criteria for choosing once
+
+## Article Structure (Suggested)
+
+**Opening hook:**
+Henry's story — switching from Midjourney to DALL-E to SD to Flux. Each time: new prompt syntax, different strengths, workflow disruption. The "new best model" trap.
+
+**The Problem:**
+- Model FOMO is real
+- Comparison content doesn't help (features ≠ fit)
+- The hidden cost: prompt expertise is model-specific
+- Leaderboards measure benchmarks, not workflows
+
+**The Reframe:**
+"Best" is contextual. What matters:
+- Your use case (photorealism? illustration? consistency?)
+- Your workflow (API? UI? local?)
+- Your prompt investment (switching = starting over)
+
+**The Framework (brief, not exhaustive):**
+
+| If you need... | Consider... | Why |
+|----------------|-------------|-----|
+| Photorealism | Flux 2.0 or Imagen 4 | Best at realistic faces, lighting |
+| Artistic styles | SDXL | Style keywords actually work |
+| Text in images | Seedream 4.0 | Only one that handles typography |
+| Image editing | Gemini/Nano Banana | Built for transformation, not generation |
+
+**The Real Lesson:**
+Pick based on primary use case. Ignore "overall best." Master one model's prompt language. The productivity gain from expertise > marginal quality difference between models.
+
+**Closing:**
+Henry's current choice (Flux for his workflow). Not because it's "best" — because he knows it. That knowledge compounds.
+
+"Go pick one. Then go build something."
+
+## Requirements
+
+**Content type:** Opinion piece with practical framework
+**Target length:** 1500-2000 words
+**Tone:** Henry voice — direct, experienced, slightly contrarian
+
+**Must include:**
+- Personal story of model-hopping (opening)
+- Specific pain points (prompts breaking, relearning)
+- Brief model overview (NOT exhaustive comparison)
+- Decision framework (table or clear criteria)
+- Clear recommendation approach
+- "I remember when..." moment (tech evolution perspective)
+
+**Must NOT include:**
+- Exhaustive model comparison (not a listicle)
+- Detailed prompt examples for each model (separate content)
+- Pricing comparison (changes too fast)
+- "Best overall" claim
+
+**Code/visuals:**
+- No code needed (opinion piece)
+- 1 comparison table
+- Hero image: abstract "choice/decision" visual
+- Optional: 1-2 example outputs showing model differences
+
+## Success Criteria
+
+**Engagement:**
+- Resonates with developers who feel model fatigue
+- Gets shared as "finally someone said it"
+- Positions Henry as pragmatic voice
+
+**SEO:**
+- Ranks for "best ai image generator" queries (contrarian angle still matches intent)
+- Long-tail: "how to choose ai image model"
+
+**Brand building:**
+- Establishes Henry's expertise in AI image tooling
+- Sets up future content (model-specific tutorials, workflow content)
+- Warm-up for Banatie content later
+
+## Distribution
+
+**Primary:** Dev.to (canonical)
+**Secondary:** Hashnode (cross-post), LinkedIn (snippet + link)
+**Potential:** IndieHackers (fits "technical opinion" format)
+
+**Social snippets:**
+
+*For X/Twitter:*
+"Spent 6 months hopping between AI image models. Midjourney → DALL-E → SD → Flux.
+
+Every switch = relearning prompts, broken workflows.
+
+The lesson: The 'best' model is the one you actually learn to use.
+
+Stop comparing. Start mastering."
+
+*For LinkedIn:*
+"Hot take: Reading 'Top 10 AI Image Generator' articles is procrastination.
+
+After 12 years of dev work and too many tool switches, here's what I've learned about AI image models:
+
+The quality difference between top models is marginal. The productivity difference between 'tried it once' and 'mastered the prompts' is massive.
+
+Pick one. Learn it. Build things."
+
+---
+
+# Review Chat
+
+(empty — article not yet written)