> [!info]- meta > **Author**: [[Alexandre Kantjas]] > **Full Title**: Alexandre Kantjas’ Post on Automation, Workflow, and Agent Heuristics > **Category**: #articles > **Tags**: #ai > **Summary**: Alexandre Kantjas provides a framework to help decide between automation, AI workflows, and AI agents based on three key pairs: decision-making and human involvement, data structure and adaptability, and reliability and risk tolerance. He emphasizes starting with automation and only moving to more complex solutions when necessary. Kantjas warns against the hype of AI agents, stating that many processes can be effectively automated without them. > > **Source**: [Original URL](https://www.linkedin.com/posts/akantjas_automation-ai-workflow-or-ai-agent-to-activity-7318265649863045120-TKTL?utm_source=share&utm_medium=member_ios&rcm=ACoAAAK2eMcBGE3emCSS0NyUYAcGgnvtx38fnXg) ## 🔦 Highlights & Commentary - 𝐏𝐚𝐢𝐫 #3: 𝐑𝐞𝐥𝐢𝐚𝐛𝐢𝐥𝐢𝐭𝐲 ↔️ 𝐑𝐢𝐬𝐤 𝐓𝐨𝐥𝐞𝐫𝐚𝐧𝐜𝐞 a.k.a how predictable the outcomes must be - and what level of risk is acceptable: ([View Highlight](https://read.readwise.io/read/01jsemh8sba3ra4t1ny8mf3e7a)) - Note: This tracks with Doctorow’s take as well, a la the 2x2 matrix of usefulness and risk. -  ([View Highlight](https://read.readwise.io/read/01jsemmqqcnfteyqwkhzz8bc7q))