Brandon Checketts

Saas companies are used to having 80-90% gross margins. I have some with 98%! I think AI is going to compress that a lot with some significant percentage eventually going to AI inference My prediction is Saas companies in the next few years see more like 30-50% gross margins

Curious how @MickeyViews interprets this as related to slow expansion at WDW in recent years

Good explanations about the Disney-DeSantis Debacle and it provides some explanation for why Disney has been playing catch-up for a few years now in expansion plans floridapolitics.com/archives/79242…

@cathrynlavery Guess I need to try out paperclip

Wow this guy is crazy. You won’t believe what he’s doing to keep his Claude agents alive 😮

Why do flights announce "non stop service" to a destination? You're flying at hundreds of miles let hour. Of course there are no stops!

I don't understand why its considered secure to be able to stay logged into my gmail account for months at a time with having to re-authenticate. But every other service logs me out after a hour or two and needs to verify (usually through gmail!) that I'm still me!!!

@itsdevdaniel That makes sense. But my apps don't have free credits, so I landed on the "warming" email account theory

@bchecketts One of the most common reasons is if you have some free credits when you sign up. Then it’s also easier on your side because you don’t need multiple inboxes Sign up again, put some dots in the email address and bam! more credits!

If you run a SaaS, you've probably seen weird signups with emails like f.o.o.b.a.r@gmail.com Gmail ignores dots in the username. So foobar@gmail.com, foo.bar@gmail.com, and f.o.o.b.a.r@gmail.com all go to the same inbox. I still haven't figured out quite why I see so many of them. I think its something with warming up the email addresses to make them seem used

The HTTP Protocol allows multiple websites to be served on a single IP Address. If using AWS Application Load Balancers, you create "Rules" that can be based on the host name to determine at which instances to point. your traffic On a webserver, it can also have multiple virtual hosts configured. Your Apache or Nginx configuration can be configured with different hostnames serving traffic from different directories. There usually is a default that is served if it receives traffic from a hostname that is not configured

AWS Route 53 has a handy feature for referring to AWS Services, even though the underlying service may change IPs at any time For example, Application Load Balancers have two IPs in two different availability zones. Amazon can change those at any time for whatever reason So in Route53, you can use an "Alias" to point to a Load Balancer and AWS will update the actual DNS records as needed so that it always points at your load balancer So never point your hostname at your ALB IP Addresses directly, because they will change

@levie It's like JavaScript libraries 10 years ago

It’s remarkable how often you need to be dramatically upgrading your AI architecture given the pace of progress in AI models right now. If you’re building agents, you basically need to throw away large parts of previous work that you setup to compensate for model limitations every few quarters. The systems you built to mitigate context window limits aren’t useful anymore, and for many use-cases it’s easier just to throw more compute at a problem today in ways that wouldn’t have worked previously. If you’re deploying agents in a workflow, you likely need to equally be rethinking your core systems at about that same frequency. The way you would deploy agents in an enterprise 18 months ago is entirely different from the best practices that you’d have today. This is partly why everyone’s working so hard right now. Right as a best practice is solidified, models improve dramatically, and that old work is rendered obsolete. Unclear that this lets up anytime soon, which is why the it pays to be so wired in right now.

Its only a matter of time until AI providers start adjusting pricing by time of day. Already OpenAI has "flex" pricing for queries that can be slower. Several offer Batch processing that can be run on the providers schedule The cost to serve these requests are based largely on electricity those rates vary by time-of-day. And increasingly, power seems like it will be the limiting factor to growth

This looks cool

Main thing I was using it for was a Slack conversation that modified files on a dev server. It was basically a way to interact with claude code, but in Slack With change to pricing, I'm planning to switch to teaching the project owner to just SSH in and resume a Claude Code session in tmux

Shut down my OpenClaw. I don't see the hype in it right now. Sure, it can do cool things, but for now I don't even need it. If you use it, what do you use it for?

Claude is on fire today!

@MerchantWords Fetching the Bluetooth Speaker example (B0DCCSFMFQ) is timing out fetching the product from Amazon. Waited 100s and still going

Something NEW this way comes. 👀 Nothing is the same as it used to be… and that’s the point. We rebuilt Power Listing from the ground up. Preview it here → buff.ly/L4EpiLS #MerchantWords #PowerListing #AmazonFBA #AmazonSeller #AICopywriting #AIWriter #AIWriting

I'm glad they took my previous recommendation about this and added your score to the Photo

The revamped Buzz Lightyear ride is tons of fun

Data Centers in space sounds like a brilliant idea Except how do you protect them from adversarial companies/nations. If we get to the point where we rely on them, it seems too big and obvious of a target for enemies to ignore

Spent a while with OpenClaw today after Anthropic nerfed Claude Pro subscriptions Grok models were as expensive and hallucinated badly Kimi 2.5 seems reasonably smart and hopefully cost efficient I created a screen session running Claude Code that I authenticated directly. OpenClaw attaches to that and uses the Claude Code session Seems to be working okay for now. We'll see in a day or two if it keeps up and costs are acceptable