Dreamforce is three weeks out. I’ve been reading the session catalog the way I read most agendas now: looking for one specific conversation.
I can’t find it.
Here’s what I found instead. Agents. Wall to wall.
Agents that book meetings. Agents that close cases. Agents that draft replies, route tickets, update records, kick off workflows. Sessions on building them, governing them, scaling them, trusting them.
The verb of this conference is act.
Every session starts in the same place: the application layer. What can the agent do? How fast? How safe? How many at once?
That’s a real conversation. I’m not dismissing it. Action is what makes a demo land, and Moscone runs on demos.
But every one of those sessions rests on the same assumption. That assumption is carrying more weight than anyone’s giving it credit for.
The assumption is that the systems underneath are understood.
An agent doesn’t act on anything. It acts on your Salesforce org. Your Jira board. Your ServiceNow queue. Systems that change constantly, in small ways, all week long. A flow edited. A permission set updated. A deploy that went out clean and passed every check.
Then it goes sideways. Cases come in. Someone pulls up the week and starts asking what happened.
Here’s the question no session starts with: can anything in the room explain what connected Thursday to Tuesday?
Not detect it. Your monitoring did its job. Explain it. Point at the change, point at the cases, and show they’re the same event.
That question doesn’t have a track.
I’ll say the quiet part out loud.
Dreamforce will answer “what can an agent do” a thousand ways across three days. That’s the conversation the entire ecosystem is built to have this year.
“Can anything explain what your systems did last week” won’t get a keynote. It won’t get a booth. It’s the conversation that starts one layer down from every demo, and almost nobody will start there.
The first conversation is about speed. The second is about whether the thing moving fast is understood at all. They aren’t rivals. One sits underneath the other.
So that’s what we are bringing to Moscone. A question. When the sessions go deep on what agents will do, we’ll be asking what can explain the systems they’re doing it to.
If a session touches it, we’ll be in the room. If a hallway conversation opens it, we’ll cancel whatever’s next.
Agents act. The agenda settles that. Whether anything can explain what they’re acting on is not settled. That’s the conversation worth three days at Moscone.
Teams heading into Dreamforce with agents on the roadmap and a Tuesday-to-Thursday question nobody can answer are exactly who we’re thinking about.
If that’s your team, we offer a complimentary Investigation Cost Audit. Forty-five minutes. Structured diagnostic across five dimensions. You leave with a scorecard that quantifies what the investigation is actually costing you: in time, in recurrence, in the reconstruction your team runs while the ecosystem demos what’s next.