Salesforce bets on storytelling over engineering

Salesforce freezes engineering, hires more sales reps, and focus on selling the story of AI to their customers

The CEO of Salesforce recently said that they won’t add more developers in 2025. This isn’t a buried lede; the title of the podcast episode that originated this line is literally “Marc Benioff, Salesforce Founder: Why Salesforce Isn't Hiring Software Engineers”. Instead, they’re looking at increasing the number of sales reps they have by thousands of headcount.

Transcript from Salesforce Ben

We’re not adding any more software engineers next year because we have increased the productivity this year with Agentforce and with other AI technology that we’re using for engineering teams by more than 30% – to the point where our engineering velocity is incredible. I can’t believe what we’re achieving in engineering.

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.

First off: many people interpreted this to mean “we will not hire a single software engineer.” This is demonstrably false! Navigate to Salesforce’s careers page and search for “software engineer”. They have over 700 open software engineering positions right now worldwide. Salesforce is obviously hiring software engineers in 2025. Benioff means that the engineering organization will maintain its current headcount. People did not focus enough on the line, “we really need to explain to people exactly the value that we can achieve with AI,” which is what the rest of this newsletter is about.

Instead of developers, Salesforce wants to hire thousands of additional sales representatives. To sell what? Their newest chat product named Agentforce. This appears to be a LLM chatbot that can answer typical frontline support questions. It can additionally act as an AI agent to perform work to your Salesforce installation.

The site Salesforce Ben has an in-depth review, where a Salesforce expert asks Agentforce six questions and rates the responses. Agentforce failed the test overall, scoring 28/60. It even failed to answer a simple pricing question. The author also notes that Agentforce could cause inexperienced Salesforce developers to incur unnecessary tech debt. However, the author liked some specific technical answers that it gave.

When it comes to Salesforce Help, Agentforce shines in specific scenarios. It performs much better when you provide clear and detailed instructions. However, I found its technical guidance to be quite high-level. While this is great for experienced users, it might leave newcomers feeling a bit lost. Personally, I found Agentforce particularly helpful for crafting validation rules – it really stood out there.

That said, I still prefer resources like the Salesforce Answers community or blogs. Advice from real-world experts often includes practical examples, clear instructions, and those helpful ‘gotchas’ that make a big difference when solving problems.

Okay, so Salesforce’s AI story is functional but has rough edges. Doesn’t it sound like they need engineering to make this better? What’s going on here? Let’s start from scratch. Imagine you just heard the sentence “Salesforce’s big bet in 2025 is their new chatbot/AI agent” what are all of the organizational changes that you might expect to see?

• If it really handles frontline support questions, then Salesforce should reduce the number of support engineers and call center staff. This appears to be their plan.

• I would expect them to add and retrain sales staff to sell the new capabilities. This appears to be happening.

• If their AI technology has rough edges and they are doubling down on selling the AI, they should gradually increase engineering headcount in AI-specific areas, since adding tech capabilities will allow them to sell the customer on even more value.

• If you had some magical 30% efficiency boost that your competitors did not have access to, then the value of adding engineers is much higher for you than it is for your competitors. So you have an incentive to spend for top talent because you can squeeze even more performance from them.

But this isn’t what is happening. Instead they’re freezing their engineering headcount. What gives? We’ve spoken before about how engineers view their companies from a bottom-up perspective, and top-down decisions often don’t make sense to them. Imagine that you work at a company called “YourCo” and your company signs a contract for Agentforce. You’re invited to several training sessions and spend a bunch of time reviewing code during those sessions, while occasionally trying the examples. You find that it is pretty handy for performing some subset of your job, but mostly breaks even or wastes time. It’s mediocre. You learn from your manager that YourCo is so large that it’s going to pay over $300,000/year for this mediocre software, and your team is getting pressure directly from executives that people working on the integration need to incorporate it into their work. And you want to scream, “with that kind of money we could have just hired someone!” How does this kind of thing happen?

Let’s look at the same scenario from the top-down view:

• Investors in your company want a positive return on their money. They also know that AI is disrupting many industries. Investors do not care about your company except to the extent that it makes them money. So they will only keep their money in YourCo if they believe that YourCo will survive the AI technology wave.

• YourCo’s board, president, and CEO are all aware of this dynamic. They are also aware that the company’s compensation scheme only works as long as the investors don’t pull their money. So they work with other executives to translate the desire to “do AI” into concrete objectives like “accelerate engineering velocity using generative coding techniques” and “look for opportunies to get rid of people increase efficiency using AI agents.”

• Salesforce’s sales reps know that companies feel this pressure, so they reach out with a compelling sales pitch: “Agentforce — and other tools that we’re going to gloss over — made our engineers 30% more efficient. They’re so efficient that we’re not even going to hire any more of them. I mean, holy shit right? Aren’t engineers so important usually? By the way, why don’t YOU want to be 30% more efficient?”

• Your executives see that this aligns nicely with your company’s need to “do AI”, so YourCo inks a deal with a volume discount. The Salesforce team hosts training sessions. Employees are generally nice and positive. And then people mostly don’t incorporate it into their workflow, because people mostly don’t incorporate ANYTHING into their workflow unless they must.

• Middle management, getting pressure to make the Agentforce deal a winner, start trickling down encouragement for people to use it.

• Feedback trickles up that people generally don’t like using it and the examples from training don’t apply to their work, but they have found it helpful for very mechanical tasks like generating filters. And it did save some people from making some support calls.

• At the next quarterly earnings call, the CEO monotones “and we have realized efficiencies in our Salesforce integration using Agentforce to automate tedious and mechanical tasks like filter generation, and quickly resolving questions without interacting with Salesforce support at all.”

• The investors are sated. For now.

You’ll note that the real value that Salesforce provides companies is not engineering. Well sure, they need to have some engineering, and that engineering does need to do something. But the engineering is just a pretext for the real value: helping companies build a story that they are catching the next wave of innovation and are not being left behind.

So let’s go back to a section from Marc Benioff’s statement:

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.

Now this is super clear, right? Everyone knows that Salesforce is going to sell a functional product at a shocking price point. They didn’t get to build that tower in San Francisco for no reason. But the AI agent is not the value that Salesforce will provide. Instead, the value is that companies that use the AI agent will get a story to tell, and they will get to flex their muscle of incorporating AI techniques into their company in case something good actually comes along.