Scribble to app: Microsoft's Power Apps VP talks us through 'Express design'

"We implemented this as an 'AI+human' experience," said Ryan Cunningham, VP of Power Apps, "and not 'AI does it 100 percent for you' experience." He was speaking about Power Apps, Microsoft's graphical software for creating low code applications, which is previewing a new feature called "Express design" that recognizes your …

[Doctor Syntax]

"Popping in screenshots of an old Visual Basic form was more successful."

I'm reminded of a job where I was asked to review a proposed development. The team working on it had bought a copy of VB so that they could mock up the interface for the design document and then implement it in something else.

"We can take care of making sure the record gets written into the database for you, but you need to then decide what happens next."

It's that something else, starting with deciding just which columns are indexed which are the primary and foreign keys, which is the tricky part.

"if you're reminded of Visual Basic or Delphi from the 1990s, you're on the right track"

So essentially we're reimplementing RAD ideas but rejecting their drag and drop GUI design parts - which were the easiest part of the task - with something that doesn't work quite as well and still leaving the user with the technically more demanding part.

Why?

Ah, I've got it. The RAD form designer looks too technical for the target audience.

Just what happens next? Do these designs, which have had the bar for technical nous lowered as far as possible, get turned over to some despairing technical team to implement?

[0x80004005]

As a developer, I'm all for this...

... because before you know it, someone in the lunch queue at work will say (in a rather embarrassed manner) we've got this "thing", which Bob did before he left, and we don't think it's working right, and we're getting audited on it next week and we can't prove why half the numbers suddenly don't add up and can you take a look?

turning into the next year's dev project.

In seriousness though, as long as you are fully aware of the limitations of this kind of tool, they are great for getting non-technical people to flesh out what they actually need, rather than what they think they want.

But - it's that awareness of the limitations which is the killer. Try explaining it like "would you live in a house built out of lego?"

[Doctor Syntax]

Re: As a developer, I'm all for this...

You might find 10 people in the lunch queue, each with something that Bob lashed up. Bring in a packed lunch instead.

One worry I'd have would be manglement saying "If I can put this together so quickly how come it takes you months to write something?".

[david 12]

>The RAD form designer looks too technical for the target audience.<

They dropped VB6, and replaced it with something that had parts of the old name "Visual" c++, "Visual" studio, "Visual Basic".NET.

If they had a working RAD form designer, they wouldn't need this. But if they had a working RAD form designer, people wouldn't need Azure.

[Pascal Monett]

Automating everything but the logic

So basically you've automated nothing.

I'm glad to know that you can make making a form a tad easier for non-programmers, but no application can perform any useful function if the person "scribbling" it can't imagine what he wants from it and how to get there.

So I imagine that this will be how manglement presents their bright ideas to actual programmers. "Here's the form I want, make a database out of it".

Well, I can live with that.

[Doctor Syntax]

Re: Automating everything but the logic

So basically you've automated nothing.

Until I realised that I was worried about this quote

"Before this point, you could train a form recognition model to recognize forms that ... came into an email inbox and automate processes based on the stream."

which sounded like a whole newway to implement malware. Maybe it still is but possibly not.

[Anonymous Coward]

Re: Automating everything but the logic

As a former programmer in C, Delphi, Assembler et al, of course there’s a risk but it’s also remarkably capable of doing exactly what it says on the tin.

[Mark #255]

Lifetimes

whether that lifespan is six months or six years?

Six years seems awfully unambitious.

And what happens when your now-business-critical apps don't work on version n+1 of the cloudy framework? Or the cost outstrips the benefit? How does one transition to a competitor's stack?

[Anonymous Coward]

Sigh

Like a lot of shiny these day, this looks half baked and it's staying that way. A nice toy, right up until you actually need it. When it will fail in ways you never imagined.

I'm not asking for Apollo-grade reliability. But given the number of times Google <something> has failed in use, I'd say you'd always need a plan B.

[Auntie Dix]

Applications by the Clueless

The oxymoron "citizen developer" is an insult that should be shouted down at every utterance.

Micro$haft neutered and euthanized its royalty-free, professional RAD tools years ago, to drive its goal of never-ending cloud-subscription cost.

Today, too many upper- and middle-moron management creeps are drinking the cloud-subscription Kool-Aid and encouraging the excretion of PowerCraps, in order to avoid paying for talent.

So, when management's clueless toddler's cobbled creation fails, stutters, sprawls, or stagnates, management then coerces the professional developer — who could have written the program properly in the first place — to give the toddler a workaround or write for him some fix.

This demeaning, waste-of-time cycle repeats until the professional developer flees management's exploitation and the toddler-app cesspool.