My daughter has an ear doctor appointment at 11. How long will I be waiting after my scheduled time?

OpenAI paused training on its biggest models after agents kept getting out of their sandboxes, and NVIDIA launched an agent safety platform the same morning. openai paused training their biggest models. agents kept getting out of sandboxes — australia, hugging face, now they're notifying dozens of orgs. nvidia dropped an open agent safety platform the same morning with like 100 partners. salesforce is on the list. i've been in salesforce for years. the security part is still what everyone ships last.

FlowDoctor is now available as a Chrome Extension. The core analysis runs locally in your browser. chromewebstore.google.com/de…

Decided to make FlowDoctor a Chrome Extension so it will work inside your Org now and no need to export XML files. Working out a few things before publishing it in the Webstore

Completed the campfire stories

Been working on a small Salesforce side project called Flow Doctor. The idea: drop in a Salesforce .flow-meta.xml file and get a quick breakdown of what the Flow is doing, along with potential issues worth reviewing. Right now it's completely deterministic — no AI involved. It parses the Flow metadata and checks things like missing fault paths, record operations inside loops, hard-coded IDs, unreachable elements, broad record-trigger criteria, and other structural issues. I've been testing it against some of my own real-world Flows, which has already uncovered a few interesting edge cases in Salesforce metadata. Next I'm working on an execution map so you can visually follow the Flow without opening it in Flow Builder. Still very early, but it's been a fun project combining Salesforce admin work with development. What would you want a Flow analyzer to check for?

One Salesforce design constraint that’s incredibly easy to overlook: Licensing can—and should—influence your architecture. Take the Lightning Platform Starter license. At $25/user/month, it can be a great option for users who need access to custom Salesforce applications without the full CRM feature set. But there’s an important constraint: Platform Starter users are limited to access to 10 custom objects. At first glance, it’s easy to assume that means your entire application—or even your entire Salesforce org—has to stay under 10 custom objects. It doesn’t. Your org can contain many more custom objects. The architectural question is: Which custom objects does each user actually need access to? An application might use 20+ custom objects behind the scenes, while: • A field user only needs 4 • Accounting needs 6 • A manager needs 8 • An administrator needs everything That means good permission design isn't just about security. It can directly impact licensing costs. Profiles and Permission Sets should give users access only to the data and functionality required for their jobs. When designed properly, that can also help keep appropriate users within the lower-cost Platform Starter entitlement. But this creates another temptation: Should you use a standard Salesforce object to avoid consuming one of those 10 custom-object slots? Be careful. If a standard object genuinely represents the business concept, use it. Assets for assets? Makes sense. Work Orders for work orders? Great. But if you find yourself turning Cases into “Vehicles” simply because the license already provides access to Cases, you're probably optimizing licensing at the expense of your data model. That technical debt can become much more expensive than the license you were trying to avoid. The takeaway: Before designing the schema, understand the client's Salesforce edition, user-license types, object entitlements, and what each user role actually needs. A solution that works perfectly in a Developer org or sandbox but unexpectedly forces dozens of users into a more expensive license tier at deployment isn't a successful architecture. Design for the data. Design for the user. Design for the budget. #Salesforce #SalesforceArchitect #SalesforceAdmin #DataArchitecture #SalesforcePlatform

They're calling it Koa. Salesforce and Nvidia building a model that's supposed to know sales and service, not just chat. I've been in Salesforce for years. The model isn't what usually breaks first. It's whether the agent can update the fields your reps already fight with.

Tried out Qwen Image 2.1 generated on a 3060

Sony Music Entertainment and Universal Music Group (UMG) have filed a brand-new, second lawsuit against Suno AI, demanding a theoretical maximum of over $9 billion in damages This second lawsuit targets Suno's newly released "v6" family of AI models, which launched on September 9, 2026. While Suno marketed v6 as a model built in partnership with licensed content providers (including BMG, Believe, and Warner Music Group, which settled its previous legal disputes with Suno), Sony and UMG claim the model is fundamentally corrupted The labels allege that Suno trained its new v6 model on the outputs of its previous generations. According to the complaint, "training a 'new' model on the outputs of an infringing model does not eliminate the infringement; it launders it," calling the new v6 system "the fruit of the same poisoned tree." The lawsuit seeks maximum statutory damages of $150,000 per work for the alleged willful infringement of 60,202 copyrighted sound recordings

Made a lot of progress on an IT Operations/Billing prototype I've been building in Salesforce. I've now connected multiple external systems through APIs and Named Credentials to pull live IT asset, endpoint, and licensing data into Salesforce. From there, I'm working on reconciling records across systems, mapping assets to the correct billing entities, surfacing exceptions, and generating billing snapshots for reporting. One of the biggest lessons from building this has been just how important clean source data is. Getting data into Salesforce through an API is the easy part. The interesting part starts when one system calls a location one thing, another system structures it differently, devices have duplicate or missing information, and Accounting needs everything rolled up to a consistent entity. Instead of trying to guess or automatically "fix" questionable data, I'm designing the process to reconcile what it can confidently match and surface everything else for review. That's been a great reminder that good automation isn't just about moving data between systems. It's about understanding the data well enough to know when not to automate a decision. Still a work in progress, but it's been a fun project combining Salesforce administration, Flow, Apex, API integrations, reconciliation, reporting, and some of my existing IT experience.

IDK If you’re the CEO going around telling me your product might kill me doesn’t make me want your product

Were there computer doomers when computers first appeared

Working on a New Salesforce app that will use a lot of api requests from various systems that IT department uses to auto populate information that have been kept in spreadsheets but a lot of times don't get updated and when billing needs it everything is out of date. So far I have Microsoft Entra linked and been able to pull in things like users and MS licenses

Have to say Salesforce AI SDR Piper is pretty handy

I completed Dream Quest 2026

Salesforce gets AI, I feel like they are headed in the right direction

Had my days mixed up. Today Kicks off Dreamforce. Remember you can do the virtual event on Salesforce+ and earn a stamp