SAS announced their new "App Factory" during SAS Explore... They are trying to make a big deal out of the fact that users don't have to use SAS to do what you would ostensibly do with SAS--build AI models. And the tool they've built to enable this grand achievement isn't generally available until 2024. So, you're building tools to make building/running models easier to do with R and Python, and further commoditizing the role SAS plays in the "AI" development cycle? And the product is at least 6 months away from being ready.
Setting aside the potential folly of their strategy, the fact they are announcing the product this early feels like desperation. And it feels like they are still very slow at pushing product through their development pipeline. Not going to help their "IPO" hopes in 2024.
Over/under on the time before this product gets ki-led off?
19 replies (most recent on top)
“SAS, where dreams and aspirations go to die.”
That phase so deeply resonates.
My dream was to build large software products. Naturally, those would create large revenue streams. So I thought my dreams matched SAS goals.
Over a long career in R&D, I was given three strong chances to help start new revenue streams. All three died of political causes.
On the first project, we all worked 40+ hours per week, and we grew revenues each year. But we competed with somebody else’s project, so all our staff were removed.
On the second project, most people worked 35-hour weeks. It took us nearly a decade to ship production. Meanwhile, our competitors were not working 35-hour weeks. So while we shipped software, we lost market share.
On the third project, over a two-year span, our entire team left. On a software project, it’s hard to achieve 100% turnover, but I saw it done.
So none of these projects generated significant new revenues for SAS. And my dreams died.
Most of us don’t get to live our dreams; they exist only to show us the right directions.
And the deal was fair: I got paid. And I had some excellent managers at SAS — just not on those particular projects.
I’ve worked on many software projects, both inside and outside of SAS. The well-managed projects were all pretty comparable.
But those badly-managed projects at SAS — those were awesomely, amazingly, off-the-charts bad. I never saw software managed so badly anywhere else.
"Microsoft and Apple are executing pretty efficiently these days."
That's because they actually demand every employee is productive and efficient. No doubt neither of those two has managers or directors who have "managed" so long they can't even back up their own direct reports. only at sas where dreams and aspirations go to die.
"Often worked 42-45 hours/week during my long SAS career."
Same here. The only 35-hour week I had was when I took a week of vacation time.
Often worked 42-45 hours/week during my long SAS career.
FWIW, the 35-hour work week was always a myth for me. I’m fact, in my last division at SAS (product management) it was officially a myth. We were explicitly instructed to account for 40 hours of work per week in the timesheet system.
Any idea what SAS paid for Boemska?
@2xyx+1oAOgifa "I don't blame people for the 35 hour weeks and the RFC and long lunches. We got that in leu of good pay and stock ownership."
Bingo. That was/is the deal.
I don't blame people for the 35 hour weeks and the RFC and long lunches. We got that in leu of good pay and stock ownership.
Thanks for that more complete explanation. I still feel that the biggest factor in the decline of SAS R&D was that people were not promoted for their ability to build software.
This caused a decline not only in competency and leadership, but also in communication and cooperation. People can’t communicate or cooperate on technical issues when they lack technical knowledge.
It's true that some of the decline was due to fiefdoms and egos. Some people with technical skills preferred to play politics.
And indeed it is hard to motivate a work force accustomed to long lunches and 35-hour work weeks. People who for many years see hard work go unrewarded may reasonably decide not to work hard.
Thus evolved a software company that can no longer efficiently build software. An integration that, in the old days, might have taken us one or two years, now takes three. A product the size of Viya takes eight to ten.
The tragedy is that this did not have to happen. Mature tech companies do tend to become inefficient, but that tendency can be resisted and mitigated. Microsoft and Apple are executing pretty efficiently these days.
“
In the amount of time they have spent re-engineering Viya multiple times, they could have built a completely new platform from scratch that does everything from 9.x code, while maintaining the Viya look and feel.
“
Having been involved long-term in the R&D of both Version 9 and Viya, I’m uncertain this would’ve been possible. To do so would require a level of internal communication, cooperation, fluid technical competency and tactical leadership that is uncommon within the rank and file, including most of the managers. Earlier threads have discussed this at length.
SAS, like virtually every mature tech company seems to always have enough time to “do it over“ but never enough time to do the basic research required to get it right the first time. To do what the above quote suggests, would require a level of organizational agility, internal cooperation (death of fiefdoms), suspension of egos, and toleration of bad news all the way up the chain. That bad news is wrapped up in about three decades of technical debt and obtuse workarounds just to please a certain faction of managers and customers.
The “country club culture” of long lunches at the RFC and 35-hour work weeks common from 2000-2020 didn’t help that either.
Integrating an acquisition and moving to a new architecture are both software engineering problems.
At SAS, people were not promoted for their software engineering skills. So, they can solve these problems, but only slowly.
As long as SAS had a monopoly in its market, speed of execution didn't matter. Now it does.
The reason for buying software, rather than building it, is to get to market quickly.
In January 2023, Microsoft bought half of ChatGPT. They've been integrating it into multiple products all this year.
So, the speed of the Boemska integration is certainly underwhelming. It's a good thing AI is not a competitive market :-)
Still, this is an exciting development: it increases the value of SAS.
Sticking with the theme of the OP, they announced SAS9 will be available until 2028, at least. And they are pushing a new update to it.
This sounds like an admission that SAS can't float the company on Viya, and can't make the transition to Viya they talked up a few years ago. Viya adoption simply hasn't gone the way they expected, and the longest software transition in modern history continues to limp along--probably never to complete. As much money as they've poured into R&D the last 10+ years, they can't manage to engineer a platform that effectively competes with what startups are doing at a fraction of the cost. So, they continue to drag all the technical debt of 9 with them. It's just sad to see.
I don't get it. In the amount of time they have spent re-engineering Viya multiple times, they could have built a completely new platform from scratch that does everything from 9.x code, while maintaining the Viya look and feel. It has been 10 years since we saw Viya previews internally. 8 years since it was widely released. 5 years since it was the new platform everyone had to get used to/pushed to. and here we are, still unable to move on.
SAS acquires Boemska more than 2.5 years ago. Radio silence about it till now. And now we hear something will be available sometime in 2024. That will make 3 years since the acquisition. Is this exciting news or just underwhelming news?
I wonder if 1Q24 is an initial estimate or a revised one. If the former, then I’d guess a delay of at least two quarters is baked in.
Apologies if this is a duplicate reply; the last one seemed to fall into the cyber-abyss.
I wonder if the SAS App Factory GA date of 1Q24 is an initial estimate or a revised one. If the former, then I fully expect it to slide by two quarters at least.
Three full years.
"SAS acquires Boemska to accelerate AI integration" -- January 7, 2021
https://www.sas.com/en_us/news/press-releases/2021/january/sas-acquires-boemska-to-accelerate-ai-integration-into-cloud.html
Haha! I had no idea it was related to Beomska. But, that's pretty sad if it is basically the same product, but it took them YEARS to get GA ready... assuming they even manage to put it out in 2024.
Wasn't this part of the Beomska acquisition several years ago? I suppose this is about how long it takes for an acquisition to make it a first class product. https://boemskats.com/products/appfactory/