General AI skeptic and manager reporting in.

Have had success at work, real value, real results.

Example: extracting a bunch of data from a tool we’re required to use at my company for getting a bunch of performance metrics. The data is useful but the interface is awful and it’s impossible to pull out trends and spot the real information I need from it. So, I threw Claude at it after months of dreaming of being able to better use the data. It generated for me in a few minutes all the data I could hope for in a CSV I was able to load into another tool that gave me deep insights almost immediately and allowed me to go make some different decisions I otherwise wouldn’t have.

What I did:

1. I have created and curated a set of sub-agents and commands/workflows for building things for me.

2. I used my build command, which details a workflow for refining, planning, implementing, code reviewing, testing, then conducting a final “product review” to determine if original requirements were met.

3. I then review the code myself before running it.

The code was solid (I’m also a very strong engineer and have tailored my agents and workflows to generate code I’d be comfortable with).

Another example: one of my teams went on a journey to convert one of our internal legacy frontend applications to a newer shared component library and eliminate old cruft that we inherited when we inherited the codebase.

The team was able to get this massive UI rewrite done in under two weeks, the updated code was better than the original code (it was all React to React, TypeScript to TypeScript), and we eliminated (literally) hundreds of thousands of lines of old hand-written over-abstracted code. Bundle sizes dramatically down, higher performance, more modern UX, and the thing is in production and working. Real value: faster product iteration in this now far smaller and easier-to-work-with codebase, far less technical debt, and faster builds, etc.

The team only used GitHub Copilot for this and it required a bunch of iteration and starting over with different instructions, but they got there and still managed to save a ridiculous amount of time; hand-writing the UI migration would have been one of those multi-month projects that went over schedule (I’ve seen and lived that movie many times before).

I’m still very skeptical of all the hype but I’ve seen very real, very valuable results out of this stuff.

edit: formatting

This is so cool. Also, it sounds like a cheeky plot to a zombie apocalypse or global contagion movie.

How subtle are the flavors? Unsubtle enough that an oblivious taster might ask, "Does this bread taste like grapes to anyone else here?" Or does one need guidance to search for the flavor?

I’ve only done grape so far. It’s on the verge of subtle vs unsubtle. If you’re real used to smelling yeast OR are a woman who has a strong sense of smell, you can smell it. Otherwise it’s just bread.

It’s kinda unfair how much better women were at smelling it (empirically)

Can we buy it?

I don’t think it’s legal for me to sell it (it is a genetically modified organism used in food), but I’d like to figure out if I can just give it away

Fully visible with naked eye in Kansas City. Beautiful magenta hues in the sky. My first time seeing an aurora in person.

I see it as well in southern KC. I've been to Alaska in the winter with hopes to the see the Northern Lights. Pretty awesome to see it this far south!

Not using TUIs (unfortunately). Work in a large business on services for real estate transactions. Heavy GUI applications. I’ve actually been investigating adding heavily keyboard-driven interfaces to our GUI apps with my teams because the multi-screen clicking around all our users have to do now is so incredibly slow and inefficient.

Some of my most productive use of software (in both personal and professional settings) have been with TUIs.

Even with all the hype around AI right now (and we’re working with that, too), we can’t not have more traditional UIs to keep a human in the loop when it matters the most and as a fallback for when AI misses the mark - but that doesn’t have to mean it’s all click heavy.

Disclaimer: I’m not a fan of TikTok and have many critical opinions of Ayn Rand’s philosophies, but…

This gives me vibes of a weird company takeover, a la the kinds of things that happened in Ayn Rand’s Atlas Shrugged: government-driven benefits to large corporations and investors with friends in the right places, government involvement, forced licensure of a core aspect of the product (Rearden Metal?).

I’d like to learn more about: what are some other similar instances of such a thing?

FYI: daggerjs.org is resolving to a GoDaddy domain parked page right now.

Will check out the repo linked at the end of your message.

Thanks for reporting this, jodacola, it seems there is domain parsing issue sometimes, you may refer to the alternative site: https://000712133.deployed.codepen.website/#overview/introdu...

It’s easy for me to get worked about about the things being done and allowed by this administration, but I have to wonder: will allowing these mega companies create more opportunities for scrappy upstarts to disrupt these giant, slow moving, clunky monoliths?

Probably not.

Look at Juniper specifically. In 2021, their revenue roughly broke down as 40% service provider, 35% enterprise campuses, 25% cloud. In 2025, that had shifted to 45% enterprise campuses, 30% service provider, and 25% cloud. That shift is mostly reflected by how much money they pumped into Mist, and how successful that was.

Scrappy little upstarts have a _really_ hard time selling networking equipment to service providers and enterprises, who require tons of arcane features that take a long time to build and validate. They also operate very much on reputation, and rely on training pipelines outside of their own organizations (i.e., certifications). On the SP side (and the more modern enterprise side) there's also the significant issue of integration with other IT systems. At that scale, people aren't just command line jockeys that log into a router to provision something - Comcast can't operate like that, they need well defined API integrations with their provisioning system.

It is interesting and noteworthy that HPE's interest in Juniper is mostly due to the success of Mist, which _was_ a scrappy little upstart that got purchased by Juniper in 2019 (???). Mist (as a product line) only got successful once it was backed by Juniper, a known player. They had a much, much harder time selling to big accounts before that.

However, it's not a random scrappy little upstart, it got started by very senior people from Cisco that couldn't get their vision executed at Cisco. Specifically, Bob Friday (who co-founded Airespace in 2001, which was purchased by Cisco and directly led to Cisco wireless controllers), and Sujay Hajela, who was an SVP responsible for enterprise and wireless at Cisco, having led the Meraki purchase. More than a decade later, Meraki - another upstart, I guess - still isn't aimed at much other than SMB.

That Mist made it as an newcomer is the exception to the rule and entirely due to those very specific people and their very specific contacts. I wouldn't be surprised if at all if Mist had initially been fully intended to be a spin-out from Cisco with the express purpose of folding them back in a decade later if they were successful enough, and it just so happened that they got snagged up by Juniper first.

A nit here: leaving Cisco to do a product Cisco should do itself is literally part of the cultural DNA of Cisco; it's practically what you're supposed to do. In years of working with/around Cisco, I saw people literally do startups for things that were just planned features for existing Cisco products.

Completely agree, I expressed that poorly - that Mist didn't just get rolled back into Cisco seems like an aberration given Cisco's spin-out culture, and I'd be curious to find out some day what happened for them to get scooped up by Juniper instead.

"Normal" startups in this space that aren't just spin-outs designed to come back to the mothership if they're successful are incredibly rare.

Appreciate the insights; this segment of the industry is my forte so this was educational.

I realize it’s impossible to predict what comes next, but I’m curious about analogs to this merger and what one could reasonably expect to happen over the next many years.

My philosophy is showing in that I don’t see these deals as good for competition or the market in general, so I’m (perhaps hopelessly) looking for the silver lining here.

Spot on about Mist (Mist AI). Great insight.

“We need bigger companies so smaller ones have a chance” is a weird take.

That wasn’t the take I was going for, but can see how it came off that way.

I’m opposed to these mega corps and looking (hoping) for some silver lining here that gives me some hope. Sibling comments have educated on that front.

Sure, and then the clunky monoliths buy the scrappy little disruptors and take them apart.

(Source: worked for a scrappy little disruptor that was bought out and cannibalized)

One way to avoid “corporate raiders" and hostile takeovers is for companies to be owned by employees.

Maybe, but what’s to stop “scrappy upstart” from becoming the next HPE?

We need companies owned by the people that built the company. Not by the C-level executives that are appointed by a board of billionaire lackeys, bankers, trust fund kiddies who are hellbent on flipping a profit at all costs.

Also of course more regulation, and higher corporate tax. Get rid of the stock manipulation tactic known as stock buybacks that only encourage short term growth/pump in price of stock.

Don't need to upload. I just tested this out because I didn't want to have to upload to asciinema:

  $ asciinema rec test.cast
  <do stuff in the terminal then ctrl-d>
  $ cat test.cast | svg-term --out=test.svg

And voila, no upload needed.

edit: formatting

Along these lines, I suggest throwing the letters in random order on the word boundaries for the player instead of forcing us to do it.

Dragging the letters to the word spots was the first thing I did to start visualizing, which was busy work I found no value in doing.

Otherwise, interesting and challenging!

Yeah! Or the team there prioritizes taking care of folks and maintaining boundaries between work and life by having that 3 hours of downtime during normal business hours.

Would love to hear other stories of businesses doing similar activities during sane weekday hours.

“Do it during the day without an outage.” is my approach.

It requires specific technology choices and planning ahead, but it is especially simple in the web era because they all go through load balancers which are redundant themselves.

The opposite example was Blizzard taking down WoW for half a day every week, which is an absurdly long time.

Even if they had a single server per game world instance, they could have orchestrated updates using automation to take each instance down for a few minutes instead of all of them at once for many hours.

My memory was that people generally regarded Blizzard taking down WoW for half a day each week as a mercy, as that was the only time some players had to shower, sleep and eat :-P