Why Salesforce Trailblazers Don’t Care About AI
We’ve been bombarded with AI hype from Marc Benioff and Salesforce for nearly two years now, and I can tell you that most Salesforce Trailblazers I talk to are fed up. As a blog publisher with a significant LinkedIn following, I’ve noticed engagement for Salesforce AI content is plummeting. These days, a solid DevOps piece gets way more traction.
With Dreamforce 24 on the horizon, we’re about to witness another leap in AI capabilities with Agentforce. Expect flashier demos and even grander promises. But I believe Salesforce has made a critical mistake in Trailblazer engagement, leaving most admins, business analysts, developers, and the rest of the Salesforce workforce feeling left behind. Here’s why:
Table of contents
The Economics of the Virtual Employee
Let’s not forget the massive financial stakes in the Salesforce ecosystem. It’s a major player in the commercial software world, raking in over $37 billion annually, with an estimated additional $57 billion in partner software and services revenue. Salesforce’s stock recovery from early 2023 lows was primarily due to improved operational discipline, but Benioff’s relentless AI push has certainly helped.
However, Salesforce’s projected growth is still under 10% – hardly growth-stock territory anymore. The glory days of rapid expansion are behind us, and the laws of large numbers make it tough for a company like Salesforce to significantly exceed 10% growth. That is, unless you’re Microsoft Azure serving up AI inference services. As giants like NVIDIA, Microsoft, and Google are experiencing explosive growth building AI infrastructure, Salesforce must juice up its growth with AI or risk becoming just a data container.
Enter Agentforce. With over 40 billion customer conversations in the US currently handled by human operators, Benioff claims he can help automate those interactions with AI for just $2 per conversation.
It’s a significant leap forward, if it works. Salesforce isn’t just peddling business automation software anymore; they’re selling you the workforce to go with it. The ROI for replacing humans with AI becomes compelling. Now that’s a growth company!
But here’s the rub: most Trailblazers couldn’t care less about keeping Wall Street happy. Even if they do, they struggle to relate to Salesforce’s existential need to pivot dramatically with Agentforce. Trailblazers are thinking, “What about all that Einstein stuff I learned at TrailblazerDX? Can’t I just figure out how to use Prompt Builder first? I don’t appreciate having this stuff forced on me!”
The Paywall Problem
The disconnect between an IT user community like the Trailblazers and the vendor’s goals isn’t new. But Salesforce has yanked the rug of self-learning out from under their end users with AI, and that’s where I think they’ve seriously screwed up.
As a paying Enterprise org owner for over a decade, I’ve watched nearly every new Sales and Service Cloud feature simply appear as new capabilities in the Setup menu. Developer Orgs even gave access to advanced, extra-cost add-on features like Field Service. This allowed anyone with an Enterprise license to test and train themselves on new Salesforce features. The excitement of sharing this self-learning on LinkedIn, blogs, Dreamin’ events, and Dreamforce helped create the vibrant Trailblazer community.
But I can’t access the key learning tool for Salesforce AI in my Enterprise org: Prompt Builder. I need to contact my AE, figure out my AI plan, and get the right license. It’s a pain, with lots of open-ended questions.
There’s incredible friction and cost involved in enabling the key AI learning tool in your Salesforce org. Very un-SaaS and un-Salesforce-like. As a result, almost zero Trailblazers coming to Dreamforce have Prompt Builder enabled in their org. That’s why Trailblazers don’t care about AI and will feel overwhelmed and offended by a massive push for Agentforce.
This approach not only hurts Trailblazers’ feelings but also causes actual damage by stifling an active Salesforce AI community. Take an important aspect of LLM operations: logging and observability. If we’re going to rely on prompts as key computational elements in future workflows, we need to use and iterate on their design after testing and feedback. Developing these systems requires new and different evaluation methods. But with so few end users discussing how they’re using Salesforce AI, there are no community or entrepreneurial projects around prompt observability.
In short, Salesforce’s AI strategy is alienating its core community of Trailblazers, and that’s a mistake they might come to regret.
