Big layoffs reported at ForeFlight.

[Yetti]

1 minute ago, Justin Schmidt said:

That's about 5 min in c or c++ if i am not using a library. 

The error was not GROK induced.  The hospital sent some 404s, so GROK had to add some headers.Typical Coder.   

5 minutes to write the code sure.   You are not including the testing and other time needed to deliver a functioning product.    Which would also include testing and documentation and delivery.    Child sick at school, 3 meetings, 2 long lunches, getting interrupted by some higher priority thing, and or vacation days.     I would be lucky to get the code by February.  

[Fly Boomer]

1 hour ago, N201MKTurbo said:

Just my two cents. But then I’m old school, I’ve been writing software since before most of the current developers parents were born.

My company tried the "free lunch" of offshoring some engineering projects to places with labor that cost a quarter or a tenth of what the previous developers cost.  It took three times as long to write the specifications because the new guys were not at all familiar with the software they were trying to fit their new stuff into.  And even with all the effort on design specifications, what came back was riddled with errors, and had to be fixed here.  Classic case of "write code until it's designed, run regression tests until it's coded, and then push out release after release for beta until the unwitting users (both internal and external) help get the code fixed, and the documentation written".

There is no substitute for well educated developers with years of experience with a particular product, and I can't imagine AI is going to change this.

[Justin Schmidt]

1 minute ago, Yetti said:

The error was not GROK induced.  The hospital sent some 404s, so GROK had to add some headers.Typical Coder.   

5 minutes to write the code sure.   You are not including the testing and other time needed to deliver a functioning product.    Which would also include testing and documentation and delivery.    Child sick at school, 3 meetings, 2 long lunches, getting interrupted by some higher priority thing, and or vacation days.     I would be lucky to get the code by February.  

No, that's 5 min to hand write a fairly complex parser, testing and all. It's one of those you pay for the experience. In contract business,  you don't have time to fafo

[Yetti]

8 minutes ago, Justin Schmidt said:

No, that's 5 min to hand write a fairly complex parser, testing and all. It's one of those you pay for the experience. In contract business,  you don't have time to fafo

You are assuming the contract is in place, legal has reviewed, you were not busy with another client, and I had to spend 15 minutes writing the requirements.

What you are not getting is kind of like Crystal Reports, experienced end users don't really need a developer. 

[Justin Schmidt]

3 minutes ago, Yetti said:

You are assuming the contract is in place, legal has reviewed, you were not busy with another client, and I had to spend 15 minutes writing the requirements.

What you are not getting is kind of like Crystal Reports, experienced end users don't really need a developer. 

Can't tell you the number of projects i have had to take over because of that thinking

[Justin Schmidt]

13 minutes ago, Yetti said:

You are assuming the contract is in place, legal has reviewed, you were not busy with another client, and I had to spend 15 minutes writing the requirements.

What you are not getting is kind of like Crystal Reports, experienced end users don't really need a developer. 

Funny story. Project came up that was previously worked on for 15 months,the contracting company internal and some of their india outsourcing. Quoted it as a saving project, code for a lot, close to go away quote. About what they already spent.

Well, i took it over, got rid of everything,  did proper engineering and completed in 2 months,  with automated testing, documentation and all.

Software engineering is a black art with science sprinkled in...also, like many trades it is the experience you pay for 

[Yetti]

2 minutes ago, Justin Schmidt said:

Funny story. Project came up that was previously worked on for 15 months,the contracting company internal and some of their india outsourcing. Quoted it as a saving project, code for a lot, close to go away quote. About what they already spent.

Well, i took it over, got rid of everything,  did proper engineering and completed in 2 months,  with automated testing, documentation and all.

Software engineering is a black art...also, like many trades it is the experience you pay for 

My Friend has a company with developers.    I had GROK write his Software coding AI policy in about 30 minutes and gave it to him.  He kept it in his back pocket. After a couple of months of the initial discussion the lead engineer came to him about writing AI Software coding Policy. He said he would email it.

Software Engineering is cut and paste and modify.   So ya it is more about how much they have to cut and paste from.

I recommended my friend to let go all the Jr. Developers.     A Senior Developer and AI coder of your choice is all you need. Keep the experience, Let AI do the coding part.

[Justin Schmidt]

1 minute ago, Yetti said:

Software Engineering is cut and paste and modify. 

Absolutely is not...programing,  perhaps but certainly not software engineering 

[Justin Schmidt]

2 minutes ago, Yetti said:

I recommended my friend to let go all the Jr. Developers

Might be seeing that company reach out to contract firms when the ball-oh-shit has run down hill and now back up to the management 

[Yetti]

4 minutes ago, Justin Schmidt said:

Might be seeing that company reach out to contract firms when the ball-oh-shit has run down hill and now back up to the management 

We are both Structure Coders/Consutants from back in the day with multiple CRM installs and other laundry list of project.  Really it's like our 1000th rodeo. 

Funny enough he uses a niche CRM for his customer base and was thinking about writing his own so he would not have to put up with the vendor.  If you think about trading meeting time/requirements time for internal tailor fitting the software with AI, it would probably be less time spent. 

[hammdo]

Back in the mid to late 80s, when I started software development, ‘code generators’ were all the rage. programming frameworks (anyone remember zAPP c++ or CodeGen ?) were next. Those were for ‘boilerplate’ code. AI is a lot like that. We also have products like PowerBI for ‘non-developers’ to build reports etc. SSRS was also the big BI tool…

Those folks hit our DBs directly (some of those tables were millions of rows) using no limits/where clauses and boom, deadlocks and issues. 
 

I don’t think AI will solve those issues unless it is taught to. We’ll see but management groups look at the new thing and it’s better than ‘sliced bread’…

Time will tell just how much AI ‘improves’ things for business. All depends on the tools and how/who uses them…

-Don

[Justin Schmidt]

10 minutes ago, hammdo said:

unless it is taught to

 

15 minutes ago, Yetti said:

Funny enough he uses a niche CRM for his customer base and was thinking about writing his own so he would not have to put up with the vendor.  If you think about trading meeting time/requirements time for internal tailor fitting the software with AI, it would probably be less time spent. 

I mentioned above my ceo has gone nutty on ai.

One thing i have been tasked with on the bench is working on creating a motif to qt port tool. Both from a non ai (using ai to create code to do it), or using ai for the port (both cloud and our internal model).

In short it's a disaster,  impossible to create code that will actually engineer and make decisions on the whole picture.  Proving impossible to train the ai to do it. Get massive gibberish

I used to say, shit if he's going to pay me to do bs then fine...i revise my earlier sentiments as it's just annoying and wasting my time and talents and eventually will tank the company 

[Yetti]

2 minutes ago, Justin Schmidt said:

 

I mentioned above my ceo has gone nutty on ai.

One thing i have been tasked with on the bench is working on creating a motif to qt port tool. Both from a non ai (using ai to create code to do it), or using ai for the port (both cloud and our internal model).

In short it's a disaster,  impossible to create code that will actually engineer and make decisions on the whole picture.  Proving impossible to train the ai to do it. Get massive gibberish

I used to say, shit if he's going to pay me to do bs then fine...i revise my earlier sentiments as it's just annoying and wasting my time and talents and eventually will tank the company 

In some aspects you are correct about having to architect the overall system.   Which is what the Senior people do in their head.   One of the first questions is why port something when you could probably write it new in a better language in a couple of days using AI.   I would feed the original to AI and then ask for suggestions.   The biggest trick is getting the LLM to not do consensus of the internet and go into analytical mode. 

 

[Justin Schmidt]

2 minutes ago, Yetti said:

In some aspects you are correct about having to architect the overall system.   Which is what the Senior people do in their head.   One of the first questions is why port something when you could probably write it new in a better language in a couple of days using AI.   I would feed the original to AI and then ask for suggestions.   The biggest trick is getting the LLM to not do consensus of the internet and go into analytical mode. 

 

First, many remaining motif is defense which cuts any use of anything cloud,  many companies don't want their products code on cloud.

Second, doing a fairly simple test ai cannot accurately or efficiently create a qt code base ui from motif, whether cloud like anthtropic or train our own models.

Third, i can very easily and quickly do it. I can architect in my head, i do that with the estimate.  While porting i am making a million design decisions for maintainability, language correctness, security and all the other ilities. Things you cannot train ai to do. Every ai written code i have seen has major problems,  security being top, code correctness, design cohesion etc. For me not worth waisting my time on it.

Rewriting a 30+ year old code base, usually is never good idea nor in the financial wheelhouse, particularly critical code.

Another issue i see with ai, much more than in the good old days of things like stackoverflow is the stupidity of programmers.  Them using ai, no longer understand what they use and you get blobs of turds. Most couldn't walk through a code base to save their life because they have had something else do it for them. Large firms code bases are falling apart now because of it. Only going to be hilarious when i retire 

[Vance Harral]

1 hour ago, Yetti said:

I recommended my friend to let go all the Jr. Developers.     A Senior Developer and AI coder of your choice is all you need.

This attitude is winning lately, and I can't really argue with it.  But no one seems to know where the next generation of Senior Developers is going to come from, when there's no time/patience/money for Junior Devs.  That's a future problem to be solved by someone else, I guess.  In the mean time, the employment landscape for legions of young software developers is a wasteland of filing literally thousands of applications (because both the producers and the consumers of the applications are using AI, natch), who are drowning in bitterness.

On the other hand, I'm sure that 50 years ago someone claimed there eventually wouldn't be anyone with enough machine language programming skill to write compilers.  Hasn't happened yet.

[Fly Boomer]

1 hour ago, Yetti said:

One of the first questions is why port something when you could probably write it new in a better language in a couple of days using AI.

That does make a lot of sense.  The original was a million lines of code, and the AI version is only 100 lines of code.  If it duplicates the functions of the original, I'm totally in favor of it.  BTW, I'm holding you personally responsible for the outcome.

[Jackk]

Or I could just not update my foreflight version until we see where their coding goes

 

 the nav data and charts are just mostly .gov or Jepp data which is what it is

Edited yesterday at 03:19 AM by Jackk

[Yetti]

1 hour ago, Fly Boomer said:

That does make a lot of sense.  The original was a million lines of code, and the AI version is only 100 lines of code.  If it duplicates the functions of the original, I'm totally in favor of it.  BTW, I'm holding you personally responsible for the outcome.

My team did a 0 issue upgrade on a 1.2 Billion dollar /year system.  We ran 600 functional tests. It was a wonderful thing.

[Shadrach]

On 1/15/2026 at 7:09 PM, Bolter said:

There is still the Avis of EFB's...FlyQ from Seattle Avionics.  I have used this since the early days, and when testing against FF there are differences, but nothing that made me want to switch.  

I was to switching from FlyQ to FF at renewal this year but decided to avoid FF because my experience with all of aviation companies that have been bought by Arcline through the Hartzell acquisition has soured me on PE.

That being said, I have seen some buggy behavior out of FlyQ lately. Do you ever see a brief course reversal on the moving map and wide variations in depicted ground speed (35-460kts). Usually only happens for a few seconds.

 

[Yetti]

13 hours ago, Vance Harral said:

This attitude is winning lately, and I can't really argue with it.  But no one seems to know where the next generation of Senior Developers is going to come from, when there's no time/patience/money for Junior Devs.  That's a future problem to be solved by someone else, I guess.  In the mean time, the employment landscape for legions of young software developers is a wasteland of filing literally thousands of applications (because both the producers and the consumers of the applications are using AI, natch), who are drowning in bitterness.

On the other hand, I'm sure that 50 years ago someone claimed there eventually wouldn't be anyone with enough machine language programming skill to write compilers.  Hasn't happened yet.

Yep I am glad I am out of the game.    Same with law firms.     Now I just use my powers for the good of myself and my people.    Best win so far was one of my friends dads was able to distribute Christmas $ gifts to his grand kids.   The gifts were generated via some software that I had GROK write.   I used him as the beta tester.   

The other thing that GROK is good at is running Monte Carlo simulations.  So I used it to model some portfolios.   I have 3 models that are running for Q1 with the KPIs set and published to a friend.   Two weeks in to the quarter one of the models is already half way to the KPI. The last two weeks of Jan will see some of the dividends being delivered. My Q1 QBR is going to be entertaining to say the least.  

[Schllc]

17 hours ago, N201MKTurbo said:

So, I’m a professional software developer. I should be writing software right now, but I’m goofing off with Mooneyspace. 

Visual Studio keeps trying to force more and more AI stuff down our throats. Most of it is just annoying. Every now and then it will give a code suggestion  that is like it read my mind. But about 80% of the time it is just gibberish. And it gets in the way of what I need to do. It isn’t really helping.

If I go to Google and search for a code example of some syntax or function I want to use, the AI will generate example code to illustrate what I was looking for. Except, it is garbage! It is  a hodgepodge of many posts on the subject. Usually a mashup of the Stack Overflow at the top of the search results.  It often has stuff in it that was deprecated years ago. I just think that AI isn’t ready for prime time yet. Any business executive who fires all their developers with the expectation that they are going to be replaced with AI has drank the Kool Aid.

Just my two cents. But then I’m old school, I’ve been writing software since before most of the current developers parents were born.

Ai appears to be fantastic at helping one narrow down options by suggesting all the absurd and impossible options first.  It’s real gift is making something ridiculous sound plausible. 
it appears to be nothing more than a higher level tool, which in the right hands can help, and in the wrong hands does more harm than good. 
Garbage in, garbage out, it is not a panacea. Not yet anyway. 

[dkkim73]

Aside from "AI", the concern was raised that the new developers will be, well, new. Also different culture (that is a can of worms but definitely not an empty can) and likely not domain-familiar (FF used to brag about how many employees were pilots). 

My own observation is that software written for technical domains gets into problems when the product managers, engineers, and support are not adequately familiar with the domain (e.g. medicine among others). Product marketing may intentionally short you on features due to lock-in, but if you at least have people for whom the tasks are not an abstraction involved in implementation and support, it's not as bad as it could be. 

Add to this the fact that while all these EFBs are "for information only and not to be relied upon as primary navigation", the fact is that almost all of us use these in ways and contexts where bad information could be fatal. 

Add to this that grooming the P&L statement will involve shortcuts and intentional lapses. Ie. you can afford to have unhappy customers as long as too many of them don't leave, or leave soon enough. Witness the changes in the rental car industry. I reported a DB error or bug to FF where the approach and lighting overlay on the night 3D view at a local airport were laterally displaced about a 1/4 mile. I got a prompt reply reflecting understanding of the issue. I am not sure how many such issues there are, but my confidence of things being fixed and maintained in the era of "Click here to have your time wasted by a chatbot" is limited. 

All the neophiles can pile on and tell me this is the New Normal and I should just get back to digging my own shallow grave. 

 

[1980Mooney]

It should be pretty clear where the future of General Aviation is going.  This topic got over 1,400 views and 3 pages of comments in less than 48 hours.  The most interest/concern on Mooneyspace seems to be in digital avionics/software development/data integration.

All this "the sky is falling" concern about a company making internal business decisions on staffing for software development.  Everyone forgets the 2023 announcement by Amazon "Amazon lays off hundreds in its Alexa division.  Everyone thinking that it was the end.  Of course the rest of the announcement was "as it plows resources into AI".  Amazon/Alexa has not lost ground or dominance.  It has steamed ahead crushing completion.  I see figures that Amazon Alexa has 70% market share to number 2 Google Home with 20% share.  Apple can't get traction.

[N201MKTurbo]

Maybe their software department was too big. There are not a lot of new features I want or need from ForeFlight. 
 

It’s like Microsoft, their office products haven’t had many new features in years, they just keep rearranging the UI to make it more confusing. I wish they would fire the developers and leave it alone.

But they have to keep the illusion that you are getting some new value for your money to support their subscription model. 
 

With ForeFlight I know I’m getting value for my subscription, I always have current charts and that’s what I really care about.

I wonder how many pilots these days have never used a paper chart?

[midlifeflyer]

12 hours ago, N201MKTurbo said:

It’s like Microsoft, their office products haven’t had many new features in years, they just keep rearranging the UI to make it more confusing.

Well said. I customized my Word interface years ago, so it’s not a problem, but I swear they move Excel features around to hide them.