Announcing Xavvi, a Great Leap Forward for App Building

[mwieder]

they all give varying answers when a single person asks the same question twice.

My memory works the same way a lot of the time.

[paul@researchware.com]

Everyone is entitled to their own choice with regard to Livecode and Xavvi. After viewing these forum comments, I thought I would add my own.

Since the late 1980's, I have made a software tool for qualitative researchers, HyperRESEARCH. Originally in Hypercard, then Supercard, then Metacard, then Revolution, then Livecode. We deliver the software for macOS and Windows - only. No iOS, no Android, no Web version - which perhaps is missing markets we could be in. Our product is dependent on Livecode. The idea that we'd recode the > 100K lines of code to some other language is just not practical to do manually. I also view Livecode as a competitive advantage in our market. We could not deliver for both macOS and Windows if we did not have a tool that allowed fast GUI development and coding and one-click (or less so with Apple Notarization) deployment to multiple platforms.

I was skeptical about Xavvi at first, but having viewed every video, read all the text, participated in all the Q&As, I am convinced of a couple of things:
1) Xavvi will only benefit the kind of software development I currently do - a niche application for macOS and Windows - in many ways. From new widgets and libraries to using AI for UI prototyping and code development and more.
AND
2) From nearly 4 decades in directing IT in a variety of industries, I have seen and participated in a lot of rapid technological advancements. The first being the microcomputer "revolution" diminishing the control of mainframes. I recall a lot of peers saying that the PC will never impact mainframe computing ("they just are not powerful enough" or "they just are toys", etc.). Nothing I or anyone else on this forum can do will stop the next technology revolution which is the use of conversational learning systems (AI) broadly. I am seeing so many industry uses (legit ones) from associates and colleagues. I am also seeing tons of spammers/scammers hopping on the AI buzz train which is a sure way to know that AI is popular and mainstream and growing rapidly.

Now is the time for Livecode to make this adjustment to their future offering and try to incorporate AI they way they have planned into their software offering. Will it be successful? Who knows. I really hope so. I think it will be a huge win for both the company, for a lot of new software developers, and, most importantly for me, to people like me who just build apps for macOS and Windows using Livecode.

For me, a "gotcha" moment was the example of the AI Coding Assistant, where you could type:
on mouseUp
make my button a blue button
end mouseUp
and the coding assistant AI part prompt with:
I think you meant:
on mouseUp
set the backGroundColor of button 1 to blue
end mouseUp
Do you want to accept this change?

I forget syntax lots of times - and have to search the dictionary. I even forget property names (or function of command names) so finding the right one in the Dictionary can take time whether looking in the script editor documentation tab or pulling up the full dictionary. This new ability alone will personally save me time because I (almost) always know what I want to do and having the system find me the right syntax would be very helpful. If, at first, I need a Xavvi script editor open in my browser to use it and copy and paste the correct syntax to my app, that's okay. It will still be faster and easier will certainly become more integrated over time.

Kevin opened a presentation on Xavvi with a quote (paraphrasing here) that in the future there will only be two types of businesses:
1) Those that utilize AI fully in their business
and
2) Those that are out of business
While I don't think that extreme view will ever be 100% true, like PCs, like the web, like smartphones, AI is the next change. Those that successfully adopt early and well will have advantage over those that don't.

[dunbarx]

Paul.

You give me hope.

Craig

[mwieder]

Paul-

Admittedly I haven't paid any attention to xavvi and don't intend to do so, so thanks for that interesting viewpoint.
Is that video archived somewhere publicly accessible?

[paul@researchware.com]

Paul-

Admittedly I haven't paid any attention to xavvi and don't intend to do so, so thanks for that interesting viewpoint.
Is that video archived somewhere publicly accessible?

At https://livecode.com/xavvi/
There are tabs at the top and bottom of the page for Overview, Tech, Pricing, etc., in addition to all the content on the various tabs, especially see the videos on the Update tab.

[stam]

I thought I would add my own.

Thank you Paul - a balanced and cogent analysis and yes, AI-based tools seem inevitable and getting ahead of the curve like LC is doing seems wise.

I was unfortunately not able to join the webinar due as was busy at work with no option to get away - someone mentioned this was recorded but I can't see a link to this - did I miss it?


On an unrelated note, I just now saw that Heather posted in this thread regarding script widgets, I had missed it completely as it was drowned in a sea of posts bemoaning our inevitable enslavement by AI or some such

A quick glance while at work was impressive, and I guess I'll need to lose more sleep while I get my teeth into this!
In case you missed it: https://lessons.livecode.com/m/98525/l/ ... ipt-widget


S.

[mwieder]

Ah! The Updates tab. Thanks - that's what I was missing.

[FourthWorld]

Kevin opened a presentation on Xavvi with a quote (paraphrasing here) that in the future there will only be two types of businesses:
1) Those that utilize AI fully in their business
and
2) Those that are out of business

I've been out of the qual space for a while. What AI features have vendors of qualitative analysis software announced? What AI features have you announced for HyperResearch?

While I don't think that extreme view will ever be 100% true, like PCs, like the web, like smartphones, AI is the next change. Those that successfully adopt early and well will have advantage over those that don't.

Adoption is a spectrum, not a Boolean. Form factor diversity is an excellent example. Let's review that evolution as a case study in the difficulty of predicting tech trends:

When smartphones premiered there were countless articles written about the "death of the PC". Those were based on dizzying hardware sales, but few at the time bothered to think it through: everyone already had a PC, but they didn't have a smartphone because they were new.

Tellingly, none of the authors of those articles even observed themselves. None of that writing was tapped out with the edges of thumbs on a 4" screen. The "death of the PC" articles were written on PCs.

Hardware sales in an emerging market tell us only that something is new, but not how it will be used when it is no longer new, when it becomes a regular part of what we do.

To see usage patterns with form factors, Internet usage is practical metric. And sure enough, as phones went from nonexistent to selling like hotcakes, usage had an upward slope.

But then normalcy happened.

Once smartphones achieved the saturation levels of PCs, people settled into usage patterns that have pretty much stabilized for the last several years, with smartphone use slightly above the midpoint and desktop use slightly below. Sure, there are fluctuations here and there as data does, but the general trend is stable world where desktops continue to be where people spend almost half their time.

Oh, and tablets, initially touted as the reason everyone would throw away their laptop? Locked in steadily at below 3%.

There's another conversation we could have about the specifics of how computing time is used and how it represents an opportunity for LC, but that's a different topic. My only point here is that beginnings are a suboptimal place to plan endings, or even understand trajectories.

The Gartner Hype Cycle reminds us of this, again and again, just a it did with the first AI boom in the '80s with Expert Systems, and nearly every new tech before and since.

It's hard to put a stake in the sand while the sand is a swirling dust cloud.

I'm not sure Apple is a stupid company for waiting until the dust settles a bit..

After all, if the impact of generative systems is as transformative as described, why would we imagine next year will be somehow more predictable than this one?

Success for small software businesses is in verticals. I see no harm in letting the giants define the boundaries. They will anyway. It's the gaps they leave in between where small business can thrive.

If "AI" is a pervasive force, it'll be around. I can't imagine 2023 will be the last year anyone starts a business, or a new business unit.

Between where we are and wherever we may be headed, there's much work to do. Work that's being done by people. On PCs.

[dunbarx]

Richard.

Nicely done.

Do you see a joke in the Hype Cycle annotations, that smacks mightily of the opening lines of "A Tale of Two Cities?

Craig

[FourthWorld]

In a world with quantum mechanics, all times are simultaneously the best of times and the worst of times.

[richmond62]

Luckily, LC central have arranged another, more accessible, live Zoom/Youtube event for this coming Wednesday,
which should help the more disgruntled (oh, I do love that word) folk round these parts put questions to the
people up in Edinburgh in, one hopes, a combative yet respectful fashion.

I am doing my damnedest to rearrange my work schedule so I can take part.
-

-
"This thesis will also take issue with what constitutes computer literacy, whether 'computer literacy' is really a relevant or necessary outcome of using computers to deliver educational content, and why busy teachers should need to become 'computer
literate' to develop bespoke computer software for content delivery and reinforcement."

With A.I. some of these objections, questions, problems may be answered.

https://www.dropbox.com/s/aknlqyovmke61 ... f.zip?dl=0

[RogGuay]

In a world with quantum mechanics, all times are simultaneously the best of times and the worst of times.

Thank you, Richard! I love this idea and will keep it in mind forevermore.

Roger

[richmond62]

"One of the common fears by non-computer specialists is that of loss of control."

Now, let's refine or refocus that to:

One of the common fears by programmers is a loss of control.

So, the biggest question must surely be how tightly the Xavvi 'robot' [for want of a better word] can
be controlled/steered by the end-user.

Of course a lot of this boils down to how much Xavvi demonstrates goal-directed behaviour, or, put
another way, how and by what its goals are determined: by the instructions of the end-user or by
preprogrammed tendencies inwith the agent [and I think I'd quite like to call Xavvi an agent at the moment].

Of course a conflict can arise if any preprogrammed tendencies inwith the agent contradict instructions
by the end-user.

BUT, I have a feeling that these fears may, ultimately, be no more logical than those expressed by the Luddites
smashing up looms that wove rugs using punch-card programs.
-

Jacquard.loom.cards.jpg (63.04 KiB) Viewed 58121 times

-
BUT, and here's, possibly, the rub, one wonders if the Xavvi agent, unlike a load of jacquard cards, is able to
modify itself (machine learning) as it progresses . . .

As I stated in my 2004 master's thesis:

"An ‘agent’ working for educators should, ideally, exhibit goal-directed behaviour;
but to achieve the intention of the end-user (educator) rather than the agent itself.

"An agent that is goal-directed to develop, say, a recipe-book program, when the
end-user wants a program to help students grasp the finer points of chemical titration is useless."

I can see no very great reason to modify that statement today.

[stam]

From what I’ve seen it can largely be used as a macro to generate layouts and scripts to speed up development significantly, and you can then edit as per normal if you want. Other than using prefabricated layouts & boilerplate code, it can be told what to create.

Not sure what you mean by “control” though, it’s not a wild animal kept on a leash
From what I’ve seen (and I may be wrong here) underneath it all it’s livecode with low-code/no-code veneer slapped on, but still livecode, although in this case browser based.

This does not seem to be a million miles from what was talked about previously other than that “AI” has been incorporated as shown in the videos I finally found on the website. But what they show suggests these are accelerators not some mindless beast that will put an app together in its totality for you.

I suspect none of us will be able to blame AI for poor apps

[richmond62]

Not sure what you mean by “control” though, it’s not a wild animal kept on a leash

No, an agent should not be like "a wild animal", but [possibly owing to a bad diet of stories that started with Mary Shelley's Frankenstein] we all have some sort of irrational fear that an agent [which is, ultimately, a congeries of code inside a machine with an OFF switch] is going to emerge, unbidden, from inside the transistors and rip our throats out.
-

-