In this Picklist Podcast episode, I sat down with Peter, Mariel, and Tim to unpack three things that jumped out in the ecosystem: newly exposed Salesforce Flow limits, Google’s latest AI model compression news, and why Flow design patterns matter more than most teams realize. First up: Spring ’26 quietly exposed Scheduled Flow and Scheduled Path limits via the REST API. It’s not a flashy UI change, but it’s a big deal—admins can finally query the limits endpoint and see how close the org is to hitting daily scheduled interview limits, instead of finding out after something breaks. I also love the idea of someone turning this into a lightweight tool (utility bar, alerts, even a platform event) so limit visibility becomes proactive. Then we got properly nerdy on AI. Tim brought up Google’s “Turbo Quant” compression, and we talked about what faster, smaller models could mean for enterprise-scale AI like Agentforce—less memory, faster inference, and potentially cheaper, more scalable features. I also called out the bigger strategic question: Salesforce has leaned into smaller models (think ~7B parameters) and new architectures like NVIDIA’s Neotron partnership, so it’ll be a “grab the popcorn” moment watching how these approaches converge. Finally, we landed on Flow design patterns—templates for repeatable automation problems. If you’ve ever opened someone else’s Flow (or your own from six months ago), you know why consistency around naming, variable assignments, and structure is basically governance, not preference.