Current SFDC software engineer - this is just cover for a widespread engineering hiring freeze since the layoffs in January 2023. There's a ton of pressure to control costs after the near shareholder revolt that caused the layoffs.
The public nature of the announcement is certainly marketing for our AI offerings as well, but at this point I think most engineers are just worried there will be additional layoffs, in the event management cannibalizes product for short term stock gains again.
Yeah, it's just an ad for Agentforce. They have been aggressively advertising on TV lately.
One ad shows a restaurant seating customers in the rain because they aren't using AI. The other ad shows someone buying clothes that don't match because AI wasn't suggesting them. (They also serve the customer shrimp, which he doesn't like.) Looking at weather forecasts and rescheduling reservations has nothing to do with AI, and I doubt AI can do anything other than write the "we're sorry" message. (Additionally, most restaurants simply ask what food you want before preparing it, so they don't have to worry about feeding you something you don't want.) Meanwhile, choosing matching colors for multiple articles of clothing is at most a satisfiability problem, which again has nothing to do with AI. I also doubt the sales floor staff needs any help with that. There are a fixed number of SKUs and colors. If a customer says "I like this shirt, can you suggest pants", I feel like 98% of adults would be able to competently assist. They also don't get paid very much, so it's unclear what the Salesforce value add is there.
IBM also does a bunch of AI advertising and the gist is that it is being used for air traffic control. Somehow, I doubt that. If AI could replace one pilot on 10% of flights, the airlines would make millions of dollars. They all still have 2, though.
I'd love to see a case study from someone other than the AI vendor. The ideas in the TV commercials aren't even good. Would love to see "well actually we're making a ton of money thanks to these products". The reason we're not seeing it is because it isn't happening.
Looks like it. The only other explanation is they just stopped investing in other products and focused on Agents, thus no need for loads more of engineering.
No way in the world they got 30% gains. 5 maybe realistically
I recently had a refreshing conversation with a bank VP (head of AI), he said yes, they do see 30-40% improvement in "some" processes, so overall maybe 0.5-1% improvement.
So I'm betting they got 30% gains in e.g. "OCRing" and quote that, ignoring that the OCR part is 1% of an entire process chain.
Their ultimate agent product strategy feels like reducing cost of sales... by automating salespeople.
I have doubts that will ever happen, but who knows?
Sales folks are highly-compensated, so even making a few of them redundant (or making existing salespeople more efficient) would be a big win for companies (Salesforce's customers).
This hiring freeze has not stopped Salesforce from acquisitions. I suppose this means that instead of investing in their current products, they're just going to buy existing ones instead.
According to Wikipedia the companies they have purchased since 2023: Airkit.ai, Spiff, Own, PredictSpring, Tenyx.
An HN article covered this topic last month. Currently, many job postings are either attempts to attract top-tier talent at minimal compensation or, in the worst cases, entirely fictitious positions.
I really don't care generally if a business feels the need to hype things like this - but these kinds of announcements have downstream effects on salaries and job availability far outside a company like this. Other management/C-suites are gonna hear "we replaced all our engineers with AI!" and similarly slow down hiring/etc. This is already being seen throughout the industry, I am aware, but such a visible player making an announcement like this, while it almost certainly is just marketing, drives me a little crazy because it indirectly affects a lot of people.
If a company exec is stupid enough to not hire when there has been no demonstrated replacement for the engineers, they are likely going to drive the company into the ground in short order.
I don’t believe that companies would be stupid enough to kneecap their own growth because a SaaS company that has stopped innovating 5 years ago decided they are happy with their current headcount and used AI as an excuse.
It is the general insecurity about new tech although I haven't seen a single job being replaced by AI yet. And I also think the market for SWE is still very good though. And the only ones that use AI on a large scale for that matter.
Salesforce came into discussion as a CRM solution for us, but the salesman was so arrogant that he was quickly booted out. If first thought it would be nice to develop against a more modern system, but today I am glad we never jumped on that train. And I hear it devolved into the same mess that CRM systems somehow always seem to be.
I think there's a case to be made that the AI boom will require more engineers in the short term, not less. Think of the myriad of features AI capabilities open for any product, not to mention a CRM platform like salesforce.
As for the time it saves-Sure, AI saves plenty of time, but in a big company most of the time isn't spent on coding. It's spent on collaboration, code reviews, meetings, debugging, looking at production logs, using internal tools, etc.
I was at Salesforce for 4 years, and during those years the company made a massive deal about:
2020: the first AI craze, introducing “Einstein” as their name for their analytics platform, and officially changing the corporate vision to being the “No. 1 AI CRM company”.
2021: Now it’s all about “Customer 360”, i.e. account-based marketing, i.e. what basically everyone else does without such a memeable name. You wouldn’t believe the number of slide decks I had to sit through with all our little product logos orbiting this stock art character straight out of Women Laughing While Eating Salad.
2022: Never mind, now we’re betting the company on a real-time unified database called Genie, which was neither real-time nor unified (and eventually not called Genie either). Got sued for that one.
2024: AGENTS. AGENTS EVERYWHERE. WE ARE AN AGENT COMPANY NOW.
So, let’s see how this holds up in the face of the next hot thing.
Let's not forget pre-2020 when the company released "Salesforce Blockchain", "NFT Cloud" and "Salesforce Web3 Platform".
This is how they operate. Just one marketing hype moment after another. The actual product doesn't really matter (and in a lot of cases doesn't even launch). They just need to keep making pretty slide decks filled with meaningless buzzwords so their customers get distracted from the fact that they are paying $500+ per user per month for a shitty web UI on top of 4 database tables.
Don’t forget IOT Cloud back when IOT was a thing. I think I was one of the first 10 people holding that certification. I really liked the architecture but the product sort of went nowhere. After learning about data cloud I feel like a lot of that architecture (iirc it was called Thunder) wound up in data cloud.
This is the most accurate comment. This is how Salesforce works. That said there revenue continues to grow and they continue to pivot to the new trend in their marketing. The underlying business and most of the software they are actually selling remains the same just wrapped up in different and new marketing each year. It works well as a strategy.
But this IS the strategy. One thing after another, giving perception that SF is "leading" the way and enterprise customers continue to listen to the SF sales reps who just need to tell the story, so on and on it goes. :)
It sounds like one of those big companies that's just very well hooked up to the institutional money printers and the main challenge for them is making up narratives to justify the money that's flowing into their coffers unconditionally.
Yeah. I got the evil eye when I told my company Genie aka Data Cloud is nothing more than a ETL and data warehouse for the different SF clouds. The SF director at my company went on a rant on how stupid I was. It was a déjà vu moment for me. I had SF and Microsoft account reps go ballistic when I informed my bosses of my opinion of their shitty products.
This is exactly correct. This is Salesforce's 3rd or 4th rebranding of Customer Data Platform, which has never taken off. This latest positioning is targeting Snowflake.
Big orgs have plenty of people that need to justify their salaries, I mean it, I'm 100% convinced you could fire half the staff of a big organization, going heavy with the management roles and lighter on operational ones and the company would become more productive and efficient.
And the higher the people in the pyramid are the better they are at selling their ideas in their neverending quest for recognition, fat bonuses and promotions.
And once they get them, more often than not by doing pointless and even harmful work, they can then leverage it by jumping to another company.
The AI trough of trough of disillusionment is coming, and these hype-driven execs are going to feel it. There are relatively few techs that have a genuine use-case for it (copilot being an obvious example).
Salesforce will just pivot in its marketing again. They latch on to the latest trend and drop it as soon as it isn't hot any more. They have constantly done this, it is their operating procedure.
Saleforce did not really suffer when NFTs died or blockchain become uncool because by then they had a new trend to promote.
salesforce is a sales and marketing company first, tech company second. it's in their interest to create a ton of buzz and hype on whatever the current thing is and how they are that thing. Then they go on to sell a basic CRUD app that has to be customized by consultants.
Part of their sales drive. We are absolutely assaulted by their sales reps trying to sell agentforce. They wine and dine non-technical executives and push the "ai agents will do everything" story. He is just reinforcing this narrative. Nothing to see here.
The plan is also to force existing customers to re-negotiate their existing long term contracts (many made on favourable terms a few years ago) to include the agentforce and data cloud stuff.
The window of opportunity for this is now: people seem to believe the 'agents' are something really quite new and different, and we haven't yet reached the peak of the hype curve. (Whereas in the boring, backwards enterprise companies salesforce is very active in we are past the peak hype for LLMs/RAGs and its hard to sell that).
A bit ironic he says he will hire 2k+ sales staff though, considering part of their pitch for agentforce is it will handle any customer interactions, sales included.
nice, strategic reduction in workforce growth that accomplishes a similar outcome to a layoff: controlling headcount to optimize margins and account for market conditions, disguised as sales and marketing. A++
《 And then, we will have less support engineers next year because we have an agentic layer. We will have more salespeople next year because we really need to explain to people exactly the value that we can achieve with AI. So, we will probably add another 1,000 to 2,000 salespeople in the short term.” 》
I guess they are building a saleforce at the expense of everything else !
They are probably struggling to keep sales number up, they need more sales and cut costs. So makes sense to hire more sales people and freeze development… but that’s a sad story, let’s fix that sad story with AI:
AI is so good that everything can be automated, except for sales which needs a personal touch.
> we really need to explain to people exactly the value that we can achieve with AI
This feels like one of those things where you're having to explain things so hard that it should give you a clue you're duping yourself. People don't get it? We just need more salespeople to explain it more. Gotta keep those KPIs up.
The funny thing is you can’t hide how bad this shit is because at some point everybody who doesn’t have a personal assistant has to deal with customer support of some sort.
And holy fucking shit has AI made every single interaction with a company ten times more painful and time consuming. It’s the worlds most boring video game where you had to trigger the right sequence of words with some dumb robot, only then to get placed into a queue of one of five remaining humans who themselves are just reading a script.
The other day I had to beat the first barricade of AI chat to get in the human queue, and then it took them literally five hours to reply. I got a text at 1am.
There is nobody who when connected to some AI agent thinks “great this will solve my problems quickly!” It’s just “wow they figured out a new way to screw us”.
Put in AI to justify saving on support staff. Then hire more people to convince customers the AI is actually just as good as speaking to a real human...
It feels like mixed messaging about whether AI is actually good.
They never seem to talk about what happens when the the dream is ubiquitously realized and nobody is left to buy whatever it they are selling because their production expenses fund someone else's sales.
At least it will be interesting to see which pushes civilization off the cliff first, end-game capitalism ouroborosing itself or the direct impact of climate change (which is obviously also related in any case).
Benioff is very far removed from what goes on day-to-day at Salesforce. He is the marketer in chief, nothing else. This interview itself is an ad for "Agentforce". They will keep hiring engineers as normal (heck there are 104 software engineering openings on their careers site as I write this).
I also don't know how he is pulling the "AI is making our engineers 30% more effective" stat when last I checked software engineers at Salesforce weren't even allowed to use AI.
Internally they are allowed. There are internal productivity tools (IDE Plugins, APIs, Frameworks, Chat) with homegrown models (in the press) and the ability to switch to other models. The various models are integrated throughout the engineering tools and Slack.
I tried to buy Pro Tools a few years ago. Avid was using a SalesForce system, which is such a monumental piece of trash that they couldn't complete the transaction. The attempt involved weeks of back-&-forth, mind-bogglingly incompetent and defective. They ended up giving me a temporary license because they thought I had a project to complete.
SalesForce, SAP, and the other purveyors of steaming legacy enterprise excrement simply point the finger at the client when end-users can't accomplish basic tasks with their systems. And of course they're prime go-tos for government work. I don't wish unemployment on their people; but the faster their monolithic junk fades away, the better.
Everywhere I've worked at that had some kind of Salesforce integration, that integration seemed almost incomprehensibly complex or a source of endless problems, and often both. But I've never been (nor wanted to be) very closely involved with any such integrations.
Is Salesforce garbage? Is that just how CRM systems are? Is everybody just doing it wrong? What's the deal?
> Is that just how CRM systems are? Is everybody just doing it wrong? What's the deal?
These kinds of tools cover 80% of what you want to do out-of-the-box.
For the remaining 20% to build it correctly you need to either hire expensive consultants or hire in-house staff to build.
Nobody budgets properly for this, and it isn't in the sales pitch, and so that last 20% is built as horrible spaghetti code by the cheapest possible devs / consultants.
Even if you wanted to pay good salaries and hire people in-house how many great engineers want to be limited to programming in Apex on salesforce?
I've only been involved with such Salesforce integrations at one company, but based on that, I can give you my take. I'd be interested to hear others'.
One, the salesforce data changes all occur through APIs (ok) which various enterprise integration tools (Informatica, Mulesoft, etc) support (ok), but those tools typically dont support easy options for retrying a change to a specific row that causes an error. If you are updating 100 Accounts in a single Salesforce Bulk API call and "5" are busy/locked, you have to build a lot of custom error handling to notice and retry just those 5 rows. Its not part of most connectors or python libraries I've seen. Also, 3 of those errors might be fatal and 2 might be retriable but you have to learn which are which or blindly retry them all. In database terms, their API transactional semantics are not statement by statement ACID but row by row within an API request.
Second, no API or SOQL operations can pull back or process or update more than 50,000 rows.
Given those two things, unless the integration person is skilled about both error handling and testing, some of the object busy/contention failures only show up in production traffic with many jobs going on so a generic integration specialist doesn't know about these Salesforce-specific pitfalls and they are discovered after the integration goes live under strange production access patterns.
EDIT: a third issue is that most Salesforce developers are UI-centric in their thinking and training and don't have database or data modeling or data integration experience to draw on so the troubleshooting for data issues on their end tends to suffer.
Salesforce has the ability for users to define their own bespoke data models. This modeling is nearly always done by sales people who may or may not be good at sales, but are almost never good at data modeling - not an insult, simply stating that data modeling isn't their job - and so the models are almost always a mess. The problems flow from there
If I were on the Salesforce BoD I wouldn't be impressed.
Let's see Benioff put his money where his mouth is, replace some C-suite seats with AI agents. Those are the really expensive salaries, and if AI is as smart as he claims it's downright reckless and negligent not to do this.
The point of this announcement is to boost sales, not to save costs. He's selling AI hype. The only thing that matters if customers are buying the hype, not the reality of AI's limitations which I'm sure the board is well aware of.
I imagine if you demonstrate that you can have AI agents as viciously competitive COOs, CFOs, and CTOs, who never need time off, never sleep, it would be something that would set Wall Street on fire.
> Let's see Benioff put his money where his mouth is, replace some C-suite seats with AI agents.
How can you replace a job that involves stone cold deterministic thinking and copying the behavior of your peers after a 3-6 month review with a lifeless machine?
AI could largely already fill the role and fulfill all responsibilities at expected level of any C suite or management only position better right now than it could a software or operational position.
Their position is fundamentally easier to do for an AI compared to operational and labor roles. They are given data and output a decision or course of action. But since they largely aren't the ones implementing said action plan it's perfectly suitable for an AI.
C suites and execs are going to do all they can to ever avoid mentioning this though.
Right. When many jobs could be reduced to interacting with a keyboard and a camera/microphone connected to a computer, I would suggest that software engineers are not at the front of the queue to be automated away. In the AI vision we're definitely there, but we have lots of company.
For better or worse it is total dollars for IT/Software salaries vs total dollars for exec salaries. And in this case software salaries is very large number that need to be cut.
This looked a lot more like marketing than reality. If there was an article on 31 Dec 2025 that said 'Salesforce didn't hire any software engineers in 2025' I would probably have more interest in it.
> The productivity gains of 30% are probably overstated
I am doubtful as well.
I could imagine 30 percent among certain engineers for certain tasks, especially if you use a popular language with popular libraries and frameworks that are well-represented in the training dataset. I don’t know how typical of a codebase Salesforce has. They could also finetune a model on their own codebase or devote a small team of engineers to figuring out which prompts, models, etc. work best for their codebase and process. In theory, those advantages could boost it beyond what testing would typically show.
But a consistent increase of “more than” 30 percent across the whole engineering workforce seems less plausible, especially lacking details on how they measured that and uptake numbers. Edited to add: Are they even confident that their engineers are using it consistently? At this scale, that’s not a given.
I’d be interested to know whether Salesforce customers have noticed a change in the number or scope of features being announced. A change of this size seems like it should be noticeable from the outside. I’d like to hear from the engineers in particular.
Salesforce core is Java. A smattering of other languages in the mix. I left Salesforce a year ago, their main developer productivity drains had nothing to do with the code base. It's their build process where it takes a minimum of a day to get code committed, even with their git on top of perforce hack which is seriously impressive, but still a process smell, that coupled with massive overhead from when dealing with inter team dependencies and various "edicts" getting passed down from on high that blow up any planning.
In short, you could have agents that code at 2x but it would have only a small impact on deliverabkes since non-coding processes have a higher impact on velocity.
There may also be engineers working harder because of being worried about getting laid off. How could they separate those productivity increases from AI?
> The productivity gains of 30% are probably overstated
it maybe not, LLMs deliver clear value in coding tasks, but the thing is that competitors also will have gains, deliver more features, fixes and products.
I am yet to have Google, Microsoft, Amazon or Facebook suggest that they’ve achieved 30% productivity gains from AI. Those are companies at the forefront of AI investment and if they haven’t achieved that kind of increased engineering productivity, I am ready to bet money on the fact that any other company claiming so hasn’t either.
Even if 30% productivity gains are true, they are probably not because of AI. They could have fired a bunch of low performers and overworked the rest of the engineers to achieve productivity gains, but even then, I’d be very skeptical if the 30% gains would stay there long term.
Readit News
The public nature of the announcement is certainly marketing for our AI offerings as well, but at this point I think most engineers are just worried there will be additional layoffs, in the event management cannibalizes product for short term stock gains again.
These companies enjoy healthy margins. Unemployed engineers can duplicate the core functionality and offer it for less.
This definitely reads more like a “everything is going AWESOME ignore anything else you hear about us” PR-ese.
I read this as messaging to shareholders more than anyone else (and sure some word-spreading about the product).
One ad shows a restaurant seating customers in the rain because they aren't using AI. The other ad shows someone buying clothes that don't match because AI wasn't suggesting them. (They also serve the customer shrimp, which he doesn't like.) Looking at weather forecasts and rescheduling reservations has nothing to do with AI, and I doubt AI can do anything other than write the "we're sorry" message. (Additionally, most restaurants simply ask what food you want before preparing it, so they don't have to worry about feeding you something you don't want.) Meanwhile, choosing matching colors for multiple articles of clothing is at most a satisfiability problem, which again has nothing to do with AI. I also doubt the sales floor staff needs any help with that. There are a fixed number of SKUs and colors. If a customer says "I like this shirt, can you suggest pants", I feel like 98% of adults would be able to competently assist. They also don't get paid very much, so it's unclear what the Salesforce value add is there.
IBM also does a bunch of AI advertising and the gist is that it is being used for air traffic control. Somehow, I doubt that. If AI could replace one pilot on 10% of flights, the airlines would make millions of dollars. They all still have 2, though.
I'd love to see a case study from someone other than the AI vendor. The ideas in the TV commercials aren't even good. Would love to see "well actually we're making a ton of money thanks to these products". The reason we're not seeing it is because it isn't happening.
There’s other things going on though this will help spin.
There’s no way that productivity metric includes the last 2-4 months.
No way in the world they got 30% gains. 5 maybe realistically
So I'm betting they got 30% gains in e.g. "OCRing" and quote that, ignoring that the OCR part is 1% of an entire process chain.
I have doubts that will ever happen, but who knows?
Sales folks are highly-compensated, so even making a few of them redundant (or making existing salespeople more efficient) would be a big win for companies (Salesforce's customers).
According to Wikipedia the companies they have purchased since 2023: Airkit.ai, Spiff, Own, PredictSpring, Tenyx.
I’ve seen SFDC hiring managers advertising positions on LinkedIn hiring SWEs as recently as last month
I don’t believe that companies would be stupid enough to kneecap their own growth because a SaaS company that has stopped innovating 5 years ago decided they are happy with their current headcount and used AI as an excuse.
(And it's not true - if you could "replace engineers with AI" we engineers would have already done it and be relaxing while AI does our work)
Salesforce came into discussion as a CRM solution for us, but the salesman was so arrogant that he was quickly booted out. If first thought it would be nice to develop against a more modern system, but today I am glad we never jumped on that train. And I hear it devolved into the same mess that CRM systems somehow always seem to be.
As for the time it saves-Sure, AI saves plenty of time, but in a big company most of the time isn't spent on coding. It's spent on collaboration, code reviews, meetings, debugging, looking at production logs, using internal tools, etc.
Dead Comment
2020: the first AI craze, introducing “Einstein” as their name for their analytics platform, and officially changing the corporate vision to being the “No. 1 AI CRM company”.
2021: Now it’s all about “Customer 360”, i.e. account-based marketing, i.e. what basically everyone else does without such a memeable name. You wouldn’t believe the number of slide decks I had to sit through with all our little product logos orbiting this stock art character straight out of Women Laughing While Eating Salad.
2022: Never mind, now we’re betting the company on a real-time unified database called Genie, which was neither real-time nor unified (and eventually not called Genie either). Got sued for that one.
2024: AGENTS. AGENTS EVERYWHERE. WE ARE AN AGENT COMPANY NOW.
So, let’s see how this holds up in the face of the next hot thing.
This is how they operate. Just one marketing hype moment after another. The actual product doesn't really matter (and in a lot of cases doesn't even launch). They just need to keep making pretty slide decks filled with meaningless buzzwords so their customers get distracted from the fact that they are paying $500+ per user per month for a shitty web UI on top of 4 database tables.
Oof, I felt that cut in my bones and I was only a customer.
I'm surprised there hasn't been a Salesforce AR/VR.
It's telling that Benioff used to work at Oracle.
https://help.salesforce.com/s/articleView?id=sf.c360_a_data_...
And the higher the people in the pyramid are the better they are at selling their ideas in their neverending quest for recognition, fat bonuses and promotions.
And once they get them, more often than not by doing pointless and even harmful work, they can then leverage it by jumping to another company.
Gervais' principle at its finest.
Saleforce did not really suffer when NFTs died or blockchain become uncool because by then they had a new trend to promote.
The window of opportunity for this is now: people seem to believe the 'agents' are something really quite new and different, and we haven't yet reached the peak of the hype curve. (Whereas in the boring, backwards enterprise companies salesforce is very active in we are past the peak hype for LLMs/RAGs and its hard to sell that).
A bit ironic he says he will hire 2k+ sales staff though, considering part of their pitch for agentforce is it will handle any customer interactions, sales included.
I guess they are building a saleforce at the expense of everything else !
"We have to hire thousands of salespeople to sell this AI agent"
Something doesn't compute
AI is so good that everything can be automated, except for sales which needs a personal touch.
Wall Street happy, bonus for all execs, the end.
This feels like one of those things where you're having to explain things so hard that it should give you a clue you're duping yourself. People don't get it? We just need more salespeople to explain it more. Gotta keep those KPIs up.
And holy fucking shit has AI made every single interaction with a company ten times more painful and time consuming. It’s the worlds most boring video game where you had to trigger the right sequence of words with some dumb robot, only then to get placed into a queue of one of five remaining humans who themselves are just reading a script.
The other day I had to beat the first barricade of AI chat to get in the human queue, and then it took them literally five hours to reply. I got a text at 1am.
There is nobody who when connected to some AI agent thinks “great this will solve my problems quickly!” It’s just “wow they figured out a new way to screw us”.
Huh. So they are telling customers that they'll make their work more AI driven while staffing up their own human sales?
It feels like mixed messaging about whether AI is actually good.
At least it will be interesting to see which pushes civilization off the cliff first, end-game capitalism ouroborosing itself or the direct impact of climate change (which is obviously also related in any case).
Why would you need buyers or even sellers? Why even have humans in the loop at all for any business process or transaction.
My understanding from all the magazines at my dentist office is that AI will replace the whole thing.
What’s even the point of humans? It’s just AI bots jerking each other off all the way down!
…now my head hurts.
I also don't know how he is pulling the "AI is making our engineers 30% more effective" stat when last I checked software engineers at Salesforce weren't even allowed to use AI.
confirmed. Take a look at this marketing material
https://www.salesforce.com/agentforce/ai-agents-roi-calculat...
They literally advertise their AI thing under the pretense of how many people you can fire and how much money you will save.
Apparently their technology is simultaneously an AI that solves business problems as well as an AI that builds AIs that solves business problems
SalesForce, SAP, and the other purveyors of steaming legacy enterprise excrement simply point the finger at the client when end-users can't accomplish basic tasks with their systems. And of course they're prime go-tos for government work. I don't wish unemployment on their people; but the faster their monolithic junk fades away, the better.
Is Salesforce garbage? Is that just how CRM systems are? Is everybody just doing it wrong? What's the deal?
These kinds of tools cover 80% of what you want to do out-of-the-box.
For the remaining 20% to build it correctly you need to either hire expensive consultants or hire in-house staff to build.
Nobody budgets properly for this, and it isn't in the sales pitch, and so that last 20% is built as horrible spaghetti code by the cheapest possible devs / consultants.
Even if you wanted to pay good salaries and hire people in-house how many great engineers want to be limited to programming in Apex on salesforce?
One, the salesforce data changes all occur through APIs (ok) which various enterprise integration tools (Informatica, Mulesoft, etc) support (ok), but those tools typically dont support easy options for retrying a change to a specific row that causes an error. If you are updating 100 Accounts in a single Salesforce Bulk API call and "5" are busy/locked, you have to build a lot of custom error handling to notice and retry just those 5 rows. Its not part of most connectors or python libraries I've seen. Also, 3 of those errors might be fatal and 2 might be retriable but you have to learn which are which or blindly retry them all. In database terms, their API transactional semantics are not statement by statement ACID but row by row within an API request.
Second, no API or SOQL operations can pull back or process or update more than 50,000 rows.
Given those two things, unless the integration person is skilled about both error handling and testing, some of the object busy/contention failures only show up in production traffic with many jobs going on so a generic integration specialist doesn't know about these Salesforce-specific pitfalls and they are discovered after the integration goes live under strange production access patterns.
EDIT: a third issue is that most Salesforce developers are UI-centric in their thinking and training and don't have database or data modeling or data integration experience to draw on so the troubleshooting for data issues on their end tends to suffer.
Their overly complex object/row/field permissions is a hot mess. Mulesoft is limited; there is a reason why they tried to buy Informatica.
Their marketing and hype machine hurts their credibility imo.
Deleted Comment
Let's see Benioff put his money where his mouth is, replace some C-suite seats with AI agents. Those are the really expensive salaries, and if AI is as smart as he claims it's downright reckless and negligent not to do this.
AI hype is exactly what I'm talking about.
I imagine if you demonstrate that you can have AI agents as viciously competitive COOs, CFOs, and CTOs, who never need time off, never sleep, it would be something that would set Wall Street on fire.
How can you replace a job that involves stone cold deterministic thinking and copying the behavior of your peers after a 3-6 month review with a lifeless machine?
If we take at face value that AI can “replace” a software engineer, it can most certainly replace most of the managers and executives above that role.
AI could largely already fill the role and fulfill all responsibilities at expected level of any C suite or management only position better right now than it could a software or operational position.
Their position is fundamentally easier to do for an AI compared to operational and labor roles. They are given data and output a decision or course of action. But since they largely aren't the ones implementing said action plan it's perfectly suitable for an AI.
C suites and execs are going to do all they can to ever avoid mentioning this though.
2. The productivity gains of 30% are probably overstated, in a likely effort to try and sell their AI products
I am doubtful as well.
I could imagine 30 percent among certain engineers for certain tasks, especially if you use a popular language with popular libraries and frameworks that are well-represented in the training dataset. I don’t know how typical of a codebase Salesforce has. They could also finetune a model on their own codebase or devote a small team of engineers to figuring out which prompts, models, etc. work best for their codebase and process. In theory, those advantages could boost it beyond what testing would typically show.
But a consistent increase of “more than” 30 percent across the whole engineering workforce seems less plausible, especially lacking details on how they measured that and uptake numbers. Edited to add: Are they even confident that their engineers are using it consistently? At this scale, that’s not a given.
I’d be interested to know whether Salesforce customers have noticed a change in the number or scope of features being announced. A change of this size seems like it should be noticeable from the outside. I’d like to hear from the engineers in particular.
In short, you could have agents that code at 2x but it would have only a small impact on deliverabkes since non-coding processes have a higher impact on velocity.
it maybe not, LLMs deliver clear value in coding tasks, but the thing is that competitors also will have gains, deliver more features, fixes and products.
Even if 30% productivity gains are true, they are probably not because of AI. They could have fired a bunch of low performers and overworked the rest of the engineers to achieve productivity gains, but even then, I’d be very skeptical if the 30% gains would stay there long term.
Pretty old trick.