<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Model Ledger</title><link>https://modelledger.pages.dev</link><description>AI, with a little more perspective.</description><language>en</language><item><title>AI agents vs workflows: differences, examples and when to use each</title><link>https://modelledger.pages.dev/articles/ai-agents-vs-workflows/</link><guid>https://modelledger.pages.dev/articles/ai-agents-vs-workflows/</guid><description>Compare AI agents with fixed workflows using support and document-processing examples. Choose the simplest system you can test, control and recover.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>Claude usage limits: why Pro runs out and when limits reset</title><link>https://modelledger.pages.dev/articles/claude-usage-limits-reset/</link><guid>https://modelledger.pages.dev/articles/claude-usage-limits-reset/</guid><description>Understand Claude Pro session and weekly limits, where to check your reset time, and what to record before paying for more usage.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>n8n vs Make pricing: compare executions, credits and real costs</title><link>https://modelledger.pages.dev/articles/n8n-vs-make-pricing/</link><guid>https://modelledger.pages.dev/articles/n8n-vs-make-pricing/</guid><description>Compare n8n executions with Make credits using a worked example. Include AI usage, retries and maintenance before choosing an automation plan.</description><pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate></item><item><title>The best AI model is the one that passes your test.</title><link>https://modelledger.pages.dev/articles/compare-ai-models-on-your-own-work/</link><guid>https://modelledger.pages.dev/articles/compare-ai-models-on-your-own-work/</guid><description>A practical framework for comparing AI tools on real work, with a reusable scorecard and a cost-per-accepted-result calculation.</description><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate></item><item><title>Chat subscription or API: which fits the job?</title><link>https://modelledger.pages.dev/articles/chat-subscription-vs-api/</link><guid>https://modelledger.pages.dev/articles/chat-subscription-vs-api/</guid><description>Compare interactive AI tools and programmable workflows by control, billing, repeatability, and operational needs.</description><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate></item><item><title>The AI cost that matters: one accepted result.</title><link>https://modelledger.pages.dev/articles/cost-per-accepted-ai-result/</link><guid>https://modelledger.pages.dev/articles/cost-per-accepted-ai-result/</guid><description>A worked example of why the cheaper API bill can produce the more expensive workflow once acceptance and review time enter the calculation.</description><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate></item><item><title>Valid JSON is only the start of a reliable AI workflow.</title><link>https://modelledger.pages.dev/articles/structured-ai-output-checklist/</link><guid>https://modelledger.pages.dev/articles/structured-ai-output-checklist/</guid><description>A practical checklist for turning AI-generated fields into usable records: schema checks, source evidence, failure handling, and a small pilot.</description><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate></item><item><title>From customer reviews to a useful AI report.</title><link>https://modelledger.pages.dev/articles/turn-customer-reviews-into-useful-data/</link><guid>https://modelledger.pages.dev/articles/turn-customer-reviews-into-useful-data/</guid><description>A repeatable review-analysis workflow that preserves source evidence and makes uncertain classifications visible.</description><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate></item><item><title>What does an AI workflow actually cost?</title><link>https://modelledger.pages.dev/articles/understand-ai-api-costs/</link><guid>https://modelledger.pages.dev/articles/understand-ai-api-costs/</guid><description>Input tokens are only part of the bill. A clear way to estimate output, retries, tools, and human review before you scale.</description><pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate></item></channel></rss>