Five Rules for Picking an AI Model That Actually Works

Five Rules for Picking an AI Model That Actually Works

Understanding Model Selection in AI

The Context of Model Choice

  • Companies like Coinbase and Lindy are shifting towards open-source models, but this video focuses on helping individuals navigate the overwhelming choices available.
  • The recent ban of Fable highlighted the uncertainty in model availability, emphasizing the importance of not relying solely on one model for work continuity.

Importance of Work Focus Over Model Selection

  • The video stresses that selecting a model should not distract from actual productivity; understanding your work needs is crucial before choosing a model.
  • Specific tips will be provided for various models (e.g., GLM 5.2, Chat GPT), tailored to different roles such as team leads or individual contributors.

Daily Driver vs. Workhorse Models

  • A "daily driver" model should perform well across diverse tasks, while a "cheap workhorse" is suitable for familiar and repetitive jobs.
  • Before selecting a model, clarify what tasks you need it to accomplish—this is more important than the specific name of the model.

Evaluating GLM 5.2 and Other Models

  • GLM 5.2 excels at producing standard business artifacts quickly under time pressure, making it ideal for routine tasks like meeting summaries or code drafts.
  • For complex or unfamiliar tasks requiring nuanced judgment, frontier models like Claude or Chat GPT are recommended over simpler options.

Choosing Your Daily Driver

Trusting Your Tools

  • Select a daily driver that can handle messy human-centric work; personal experience with codecs shows ease of use despite its intelligence limitations.
  • Ensure your chosen harness aids productivity rather than becoming an additional distraction; consider cost-effectiveness when selecting tools.

Differentiating Task Complexity

  • Distinguish between simple artifact production (GLM style work) and complex problem-solving (fable type asks); choose models accordingly based on task complexity.

Testing and Implementing Models

Practical Testing Approach

  • Test your selected daily driver with real inputs relevant to your work before fully committing to ensure it meets your needs effectively.

Navigating Company Restrictions

  • In corporate settings where choice may be limited, evaluate available models against actual utility; advocate for better tools if current options fall short.

Optimizing Team Efficiency

Streamlining Processes

  • Small businesses should focus on essential recurring artifacts that drive customer value rather than overwhelming teams with too many models.

Specialist Tools for Specific Needs

  • Identify specialized tools (e.g., Flux for ads or LTX for video editing); prioritize understanding job requirements over memorizing tool names.

Adapting to Evolving AI Landscape

Learning from Industry Trends

  • Major companies are adapting their strategies by exploring various routing systems to optimize costs while enhancing performance through intelligent choices in AI models.

Emphasizing Individual Needs

  • There’s no one-size-fits-all solution; tailor your approach based on specific job requirements and how efficiently you can integrate these tools into workflows.

Key Takeaways from Model Selection

Understanding Task Complexity

  • For straightforward tasks with ample examples online, GLM 5.2 may outperform other options; however, complex problems necessitate advanced frontier models like Claude or Chat GPT 5.6.

Simplifying Choices

  • Avoid copying others blindly; assess how challenging the task is instead of just focusing on volume—ensure that any chosen model simplifies rather than complicates workflow processes.
Video description

Full model routing guide: https://natesnewsletter.substack.com/p/which-ai-model-to-use?r=1z4sm5&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true Every AI model suddenly looks replaceable, and picking the right one has turned into a second job. This video is a practical model picker: how to choose which AI to use for real work without overpaying or drowning in model names. My Links 🔗 👉🏻 Newsletter: https://natesnewsletter.substack.com/ 👉🏻 X: https://x.com/natebjones 👉🏻 TikTok: https://www.tiktok.com/@nate.b.jones 👉🏻 Instagram: https://www.instagram.com/nate.b.jones What's really happening when every AI model suddenly looks replaceable? The common story is that the smartest model wins. The real question is which intelligence a specific job actually needs. In this video, I share the inside scoop on how to pick the right AI for the work in front of you: - Why daily-driver models differ from cheap workhorse models like GLM 5.2 - How to route familiar work to cheaper models and review it fast - What to keep on a frontier model when the shape of the job is unclear - Where specialists win: images, video, live web, and coding harnesses The models will keep changing, but if you route by the job and keep your context portable, no single model going away can stall your work. Chapters: 00:00 Why picking an AI model suddenly got hard 01:42 Start with the job, not the model 02:28 Where GLM 5.2 fits: familiar, repeatable work 03:25 When to pay for a frontier model 03:53 Your daily driver and why the harness matters 04:47 Fable-style problems that need the strongest model 05:40 Test any model on your own work first 06:13 Using AI at work: permission comes first 06:59 Small teams: route your five recurring artifacts 07:53 Specialists for images, video, and live web 09:51 What Coinbase, Cursor, and Lindy are actually doing 12:47 Five rules for picking a model Listen to this video as a podcast. Spotify: https://open.spotify.com/show/0gkFdjd1wptEKJKLu9LbZ4 Apple Podcasts: https://podcasts.apple.com/us/podcast/ai-news-strategy-daily-with-nate-b-jones/id1877109372