Sonia Cuff walks through estimating and tracking the cost of an AI workload. She starts with a website chatbot for a hiking shop, checking how many tokens a typical customer conversation uses and how newer models use fewer tokens for the same text. The demo then moves to Cost Management in the Azure portal to see which Foundry models are driving spend, and into Microsoft Foundry for detailed token usage. Rick Claus and Sonia explain why AI costs depend on user and model behavior, not just architecture, and why unusual token consumption is a signal worth watching.
Watch the full episode of Update & Restart here:
🎙️ Speakers
Rick Claus — Principal Cloud Advocate, Microsoft
LinkedIn: https://www.linkedin.com/in/rickclaus/
Bluesky: https://bsky.app/profile/regularitguy.com
Sonia Cuff — Principal Cloud Advocate, Microsoft
LinkedIn: https://www.linkedin.com/in/soniacuff/
Bluesky: https://bsky.app/profile/soniacuff.com
Update & Restart is a practical series for IT professionals navigating how AI fits into real-world infrastructure, operations, and troubleshooting. Hosted by Rick Claus and Sonia Cuff, the series cuts through hype to show how tools like Azure Copilot and Microsoft 365 Copilot work in real environments—where security, reliability, and production constraints still matter.
Each episode combines hands-on demos, real scenarios, and practical advice to help IT pros understand what AI can actually do today, when to trust it, and how to adopt it responsibly without breaking systems or workflows.
Subscribe to Microsoft Learn on YouTube: https://aka.ms/LearnOnYouTube
Follow Microsoft Learn on social:
LinkedIn: https://www.linkedin.com/showcase/microsoftlearn/
Instagram: https://www.instagram.com/microsoftlearn/
TikTok: https://www.tiktok.com/@microsoftlearn
X: https://twitter.com/MicrosoftLearn
