For whatever reason, Salesforce has failed to capitalize on the AI excitement/craze [1]. Its earnings growth is just not what it used to be (i.e. during the peak cloud era of 2010s-202x).
A move this aggressive (e.g. pushing companies on Slack to pay 10x more, immediately, or get lost) is not isolated and probably the result of institutional forces. It's not like the random sales person in charge of this decided to be destructive. Salesforce the company is getting squeezed and this is one of the outgrowths of that pressure. And it speaks to the insane dysfunction that must be taking place in the bowels of Salesforce right now, I'm sure it's crazy.
It's really surprising -- Slack is the poster child of an app where AI-based semantic search (e.g. RAG) would be incredibly useful. Yet despite Marc Benioff's grand proclamations about AI [0, 1], you barely see any AI integration into one of Salesforce's most universally-used products.
They have AI features in Slack but they just aren’t that useful. The RAG search is the most useful one, but it falls short of solutions like Dust or Glean because it only covers a single silo (Slack). AI search is way more useful when it searches across Notion, Linear, Slack, etc so you’ll buy that instead of the Slack AI addon.
The API changes are scummy, I agree. It’ll generate some ARR short term but ultimately people will be looking elsewhere, new companies will start on alternatives and others switch when the opportunity arises. It’s also not like Slack is a beloved product.
Salesforce as a company hasn't been innovative in 20 years. It's no surprise that they can't make anything of AI outside of a couple fancy marketing campaigns.
I know a few engineers in different companies within Salesforce. They're under lots of pressure to integrate AI everywhere, and leverage it. The way they've talked about it gives me strong "flailing around desperately" vibes, when the smarter money is on making more minimal but targeted efforts, or at least waiting to see what happens the other side of the bubble.
I don't understand why Slack hasn't fully implemented LLMs. Imagine as a new comer, you don't understand why a product decision was made 3 years ago. You ask Slack to summarize the conversations on why this choice was made based on messages 3 years ago. How powerful is that?
Slack can probably charge an extra $10/month/user for this.
Has any company got this feature? Sounds like the kind of thing that sounds good in theory but is hard to actually pull off. To complete this query you'd have to process almost the entirety of the chat history in every channel. Which sounds extremely expensive, and we know LLMs start to go off the rails when you give them too much context.
Slack added AI features for something like ~$5/user/mo. Nobody got the addon because it was stupid. So Slack bundled AI and increased the base subscription by ~$5/user/mo. Nobody uses the AI features still, and we are all $5/user/mo poorer.
Source: I work at an MSP and we have a ton of clients on Slack.
I mean, they really really tried to be the low code provider. But, as far as. I'm aware, no one really likes Salesforce as a product, it's integrations are poor generally.
It's a CRM. AI won't help there, customers already hate getting harassed by cold calls and endless AI support bot loops. They are just hitting market maturity.
Can you blame him? Listening to the latest AI slop hype on twitter and elsewhere, you’d walk away thinking that LLMs have equivalent performance to humans when it comes to coding tasks. Just because it can one shot fizzbuzz or make a recipe app. (And if you disagree, you’re a hater!)
“You’re just not doing it right. Have you tried upgrading to Claude 9000 edition/writing a novels worth of guardrails/using this obscure ‘AI FIRST’ IDE/creating a Goldberg machine of agents to check the code?”
Congratulations! How did you get the process started for distributing your game? Were there any early channels for marketing you found particularly effective?
Then you will filter out bargain hunters -- they require at least as much support [1] -- and focus on price insensitive customers (they are the most profitable). Price insensitive customers also provide a clear indication of what people will willing pay for (people who want free don't want to pay for anything so there is no clear signal).
Your job is to solve your customers' problems. Except for how much your product costs.
[1]: Notice that you already have a complaint from a non-customer. You don't want customers who want you to solve that problem.
I wanted to share this power-up I've been building for Trello over the last year. It lets you sync cards between Trello boards -- a feature that Trello does not natively offer.
The existing power-ups that do this are pretty complicated to use, have unreasonably high pricing, and don't offer instant syncing in many cases. So it made sense to build something that was simple(r) and cheap(er) to use. Happy to answer any questions!
Plus a lot of people who tried to pass themselves off as Syrians aren't Syrians at all.
The asylum claim processing authorities were a bit overwhelmed, but soon learned to ask questions such as "can you tell me in which city you used to live, what was the name of the mayor, what was the closest mosque/church, your address, your school" etc. Few of the fake ones could answer such probing in a satisfying way.
Isn't the tech union the one striking? So what is he implying -- that perplexity would automate the software development of the NYT needle or something?
Meh, what I've seen is that we continually move the goalposts for AGI, and even GTP-3.5 would have been considered AGI by our standards from just 5 years ago.
Not sure why you got downvoted, this is basically the logical conclusion of programming in some sense. Sure, generating code from docs via an LLM will be riddled with bugs, but it's not like the sloppy Python code some postdoc in a biology lab writes is much better. A lot of their code gets to be correct via trial and error anyway.
"Professional" programmers won't rely on this level of abstraction, but that's similar in principle to how professional programmers don't spend their time doing data analysis with Python & pandas. i.e. the programming is an incidental inconvenience for the research analyst or data scientist or whatever and being able to generate code by just writing english docs and specs makes it much easier.
The real issue is debuggability, and in particular knowing your code is "generally" correct and not overfit on whatever specs you provided. But we are discussing a tractable problem at this point.
> Saying “we live in a society without a counterculture” sounds ridiculous the more you think about it. How could it possibly be true, especially when you consider the past? And a lot of the 14 "warning signs" are general enough that they've always been true to some extent.
> But somewhere between your 38th Marvel movie and the millionth Heard-Depp trial rehash video, you might start to believe it. Even if it isn’t new, even if it’s easy to escape, and even if it’s not that bad, a cloying sameness occupies the cultural mainstream. It seems impenetrable, same as ever. But it’s especially surprising given how much creative work today exists outside the mainstream.
> It is a jarring contrast. At no point in history have people created so much with so few channels for consuming their work. Most consumers get their content through a narrow straw — TikTok’s “For You” page, the first page of Google’s search results, Instagram’s explore tab, miscellaneous streaming sites, and so on. Many lifetimes worth of creation get aggressively filtered down into a (very optimized) stream of content.
A move this aggressive (e.g. pushing companies on Slack to pay 10x more, immediately, or get lost) is not isolated and probably the result of institutional forces. It's not like the random sales person in charge of this decided to be destructive. Salesforce the company is getting squeezed and this is one of the outgrowths of that pressure. And it speaks to the insane dysfunction that must be taking place in the bowels of Salesforce right now, I'm sure it's crazy.
[1] https://qz.com/salesforce-beats-q2-earnings-ai