Plan Forge Blog — From the Forge https://planforge.software/blog/ Lessons, patterns, and A/B test results from building with AI coding agents. By Scott Nichols. en-us Fri, 11 Apr 2026 00:00:00 GMT The A/B Test: 99 vs 44 — Same App, Same Model, Same Time https://planforge.software/blog/ab-test-plan-forge-vs-vibe-coding.html https://planforge.software/blog/ab-test-plan-forge-vs-vibe-coding.html Fri, 11 Apr 2026 00:00:00 GMT We built the same .NET app twice: once with Plan Forge guardrails, once with pure vibe coding. Same model. Same time. The quality score: 99/100 vs 44/100. 60 tests vs 13. The data speaks for itself. The 80/20 Wall: Why AI Agents Break What They Build https://planforge.software/blog/the-80-20-wall.html https://planforge.software/blog/the-80-20-wall.html Mon, 07 Apr 2026 00:00:00 GMT AI coding agents get you to 80% fast — then start breaking things trying to close the last 20%. The fix isn't a better model. It's a better methodology. Spec Kit + Plan Forge: Write the Spec, Enforce the Build https://planforge.software/blog/spec-kit-plan-forge.html https://planforge.software/blog/spec-kit-plan-forge.html Mon, 07 Apr 2026 01:00:00 GMT I started with Spec Kit and loved it. Specification-based planning and agentic coding is like peanut butter and jelly. As my ideas grew, Plan Forge was born. Here's how they work together. I Built Guardrails for AI Coding Agents — Here's What I Learned https://planforge.software/blog/guardrails-lessons-learned.html https://planforge.software/blog/guardrails-lessons-learned.html Mon, 07 Apr 2026 02:00:00 GMT After 18 guardrail files, 19 reviewer agents, and 9 tech stack presets, here are the seven hard-won lessons from building Plan Forge — and the mistakes that taught them. Quorum Mode: What Happens When 3 AI Models Review Each Other's Code https://planforge.software/blog/quorum-mode-3-models.html https://planforge.software/blog/quorum-mode-3-models.html Mon, 07 Apr 2026 03:00:00 GMT We built a feature twice — once with a single model, once with three models in parallel consensus. Both passed all gates. But the code quality difference was measurable. From WhatsApp to Shipped PR: How I Automated My Entire Dev Workflow https://planforge.software/blog/unified-system.html https://planforge.software/blog/unified-system.html Mon, 07 Apr 2026 04:00:00 GMT Three open-source tools wired into a closed-loop system. Describe a feature from your phone. The system hardens the plan, builds it, captures every decision, reviews independently, and ships. One Framework, Seven AI Agents: Why We Stopped Picking Favorites https://planforge.software/blog/seven-agents.html https://planforge.software/blog/seven-agents.html Mon, 07 Apr 2026 05:00:00 GMT Your team uses Copilot, Claude, Cursor, and Gemini. Plan Forge generates native guardrail files for all 7 agents — same quality gates, same pipeline, same 19 reviewer agents. One setup command. From Impossible to 7 Minutes: A Year of Building AI Coding Guardrails https://planforge.software/blog/the-journey-from-impossible-to-seven-minutes.html https://planforge.software/blog/the-journey-from-impossible-to-seven-minutes.html Fri, 11 Apr 2026 12:00:00 GMT A year ago, getting enterprise-grade code from an AI agent was nearly impossible. Today we build a 99/100 app in 7 minutes. This is the story of that journey.