Salesforce launched Agentforce Vibes in late 2025. It promises faster change with fewer intermediaries, but is speed enough to solve the hard problems?
How I See It
Salesforce launched Agentforce Vibes in late 2025. A moment many will call “significant.” The idea is simple: You express intent in natural language, and an AI agent turns it into Salesforce configuration, workflows, logic, and code. Less friction. Fewer intermediaries. Faster change
Why it matters
It’s an execution accelerator. This isn’t fundamentally new code generation. It is prompt-driven automation of CRM change, wrapped in conversational interfaces. If Salesforce becomes intent-driven, CRM changes from being a backlog-driven machine to a continuously adjustable instrument. For large organisations, this is an important change.
The upside
The real value is speed. Feedback loops shorten, experimentation costs drop, business teams feel closer to the system, and technical teams spend less time translating. In the environments I’ve seen, vibe coding already lowers the cost of experimentation and tightens the feedback loop between business and technical teams.
The flip side
AI does not truly “understand” the business. Vibe coding mainly reduces manual efforts and accelerates repetitive tasks. It inherits the imprecision of the inputs: Unclear assumptions, local shortcuts, and governance gaps are amplified, not fixed. At scale, these imperfections compound. In short, vibe coding just optimises delivery mechanics, not decision quality.
Vibe coding cannot replace operating models and project management practices. Weak structures simply get accelerated.
My take
Vibe coding is useful but not transformative. It is an execution accelerator, not a thinking substitute. Its success depends more on organisational discipline than the actual model itself.
So here’s the tricky question: If an organization couldn’t manage Salesforce change before, will it doing it faster with AI help? Or just make the mess more visible?