The trick is simple. Raise the price for a week or two, then relist it at the "discounted" price, which is actually just the normal price with a strikethrough next to a number that never really existed. You feel like you got a deal. You didn't.

What got me thinking about this: the products doing it best aren't obviously sketchy. Good ratings, real-looking reviews, established sellers. The fake discount is often the only tell, and most people don't check price history before buying.

Ran into something this week that's been bugging me: gave a local model the exact same prompt, replayed it twice with nothing else different, and got a correct answer once and a wrong one the second time. Not a hard edge case either a pretty basic task it clearly CAN do, it just didn't do it reliably both times.

Made me rethink something: a lot of what gets blamed on "the model isn't good enough" might actually be a consistency problem, not a capability ceiling. Those are very different things to fix capability needs a bigger/better model, consistency might just need lower temperature, better prompting, or accepting you need a verification step regardless of model size.

Curious if others building with LLMs (local or cloud, doesn't matter) have run into this do you design around inconsistency (retries, verification passes, structured output constraints), or has this mostly not been an issue for what you're building?

I work on Shopify apps, and this is something I genuinely can't resolve.

Group buying where shoppers team up so everyone unlocks a lower price is enormous in Asian ecommerce. Pinduoduo built a top-tier company on the mechanic. In the West it's essentially absent. Groupon isn't a counterexample: that was a marketplace selling daily deals, not a shopper recruiting other shoppers into a shared purchase.

One thing we ve learned from talking with businesses is that having a problem doesn t necessarily mean someone is ready to change.

A team might know its current workflow is slow or frustrating and still continue using it because it s familiar.

The more useful question for us became:

What changed recently that made solving this problem important now?

Here s my rule of thumb, and it s not scientific: if you can t remember what one of your tools does without opening it, it s not saving you time anymore. It s just another tab you feel guilty about.

I watched this happen to a client a few months ago. Eleven tools. Eleven logins, eleven bills, eleven quick trainings she sat through and then forgot. She was spending more hours managing the stack than she used to spend just doing the actual work by hand. That s the tell if you ask me. The tools were supposed to buy her time back, and instead they became a second job.

Genuine question for the room.

If an employee's work email and password have ever been part of a breach that wasn't yours, how did you first hear about it? Was it something you'd set up in advance and got an alert from, or did a customer flag something odd, or did you only work it out after an actual incident? I've also seen cases where founders only stumbled across it months later by accident.

I'm trying to get a clearer picture of how this problem actually surfaces for small teams. The version in vendor whitepapers always feels a bit too tidy compared to what I hear when I ask people directly.

Yesterday @fmerian recommended Googly Eyes from @Sindre Sorhus, which I immediately downloaded! After enjoying the eyes following my cursor around for the day, that got me thinking - what other fun, goofy apps do you all have for your Mac, Chrome, or other devices that don't necessarily serve a purpose other than to bring joy and break the monotony? @Scroll Buddy is another one that comes to mind that I have saved in a collection.

We see productivity apps, AI-based software, open-source products, etc. all the time, but what about your saved products someone made after their 9-5 or on a weekend just to have fun and test their skills?

When you're bootstrapping multiple products, there's this physical feeling that shows up and nobody ever talks about it. Your stomach is somehow empty and full at the same time. This knot that just sits there while you're trying to figure out which project needs you most.

I run Sparkum, Biteme, and LifeLines all under Onyx Labs. No investors. Every dollar is ours. Some days that's exciting. Other days it's just heavy.

A few things that actually help me:

Get specific. The "everything is overwhelming" feeling is almost never true. It's usually one or two things hiding behind everything else. Name them. The rest gets lighter.

Something we ve been learning in B2B:

Having a painful problem doesn t automatically mean someone is ready to buy a solution.

A team might complain about spreadsheets, manual attendance, messy reporting, or repetitive admin work and still keep doing it the same way.

Why?

Something I've watched break plan after plan: teams fill the sprint to capacity with planned work, then act shocked when bugs, support fires, and "quick favors" blow it apart. But that unplanned work isn't a surprise. It shows up every single sprint. The only thing you don't know is which fire, not whether there'll be one.

So you've got a plan built on 100% of capacity going to planned work, in a world where it never once has. The gap gets paid for by people quietly working late, and by the roadmap slipping in a way nobody scheduled.

The teams I've seen handle it well treat unplanned work as a first-class citizen. They reserve capacity for it on purpose, based on what it actually ate last quarter, instead of pretending this sprint will be the clean one. It never is.

For people who run teams: how do you actually model the reserve for unplanned work? A fixed percentage? Gut feel? Do you protect it, or does it get eaten the moment planned work runs long?