Dreamforce 2026: What We Are Going to Find Out

We are going to Dreamforce 2026 as builders, not to be sold to, and this post is the actual list of questions we are taking with us. We build on the Salesforce platform for a living, so the things we want answered are practical: what Agentforce and Flex Credits really cost once you run them, how mature Revenue Cloud migration is for real CPQ customers, where the AppExchange and AgentExchange quality bars actually sit, and how far MCP has spread inside the ecosystem. If you will be in San Francisco between September 12 and 18 and any of these are on your mind too, come compare notes.

I am a Salesforce engineer, and I go to Dreamforce for the hallway conversations, not the keynotes. What follows is not a preview of what will be announced, because I do not know that yet and neither does anyone selling you a prediction. It is the set of open questions I want real answers to from the people who are actually shipping.

Why go to Dreamforce as a builder with no booth?

We are visiting, not exhibiting: no booth, no session, no badge scanner. That is deliberate. The value of a conference to a team that builds on the platform is not the announcements, which we will read anyway, it is the candid answers you cannot get from a pricing page or a launch blog. What did a migration actually cost you? Where did the security review bite? What broke that the release notes did not mention? Those answers come from other practitioners in a hallway, not a stage.

So we go to listen. We spend the week comparing notes with people who have shipped the things our clients are about to attempt, and we bring what we learn back to the work. This post is the honest version of that intent: here is what I am curious about, stated as questions, so you know exactly what I am trying to find out.

What do Agentforce and Flex Credits actually cost in practice?

The first question is the one every client asks and no slide answers cleanly: what does Agentforce really cost once it is running, and how do Flex Credits behave under real usage rather than a demo? Consumption pricing is honest in principle, but the practical question is whether teams can predict their bill, and where the surprises land as usage scales.

I want to hear from people already in production: did the credit model match their forecast, what drove consumption faster than expected, and how did they keep it in check. We wrote up the Agentforce pricing model as we understand it, and Dreamforce is where I get to pressure-test that understanding against people paying the actual invoices.

How ready is Revenue Cloud migration for CPQ customers?

With Salesforce CPQ now end-of-sale, the question on every CPQ customer's mind is how mature the path to Revenue Cloud really is. Not whether it exists, but whether real teams have completed it, how long it took, and what the data-model and configuration rebuild actually involved once they were in it rather than reading about it.

I am taking the same honest framing we used in our piece on Salesforce CPQ end of life: this is a forced decision, not a forced destination. What I want from the week is ground truth on how the migration is going for the teams who have started, so we can advise our own clients from evidence rather than optimism.

How high is the AgentExchange listing quality bar?

As agents become listable products, the open question is what the quality bar looks like in practice: what does it take for an agent to be genuinely useful and safe enough to publish, rather than a demo dressed up as a product. Every new marketplace starts with a rush of listings and then sorts itself out, and I want to understand where that bar is settling.

This matters to anyone thinking about building on the platform to sell, not just to use. The answer shapes whether an agent listing is a real distribution channel worth the effort, and I would rather learn that from builders who have gone through it than guess.

What are ISVs seeing in the AppExchange security review?

For anyone building to publish, the AppExchange security review is the gate, so I want to hear what ISVs are actually experiencing in it right now: what is being scrutinized more closely, what is tripping teams up, and how long it is taking end to end. Security reviews evolve, and the lived experience is usually ahead of the documentation.

We keep a practical AppExchange security review checklist, and this is exactly the kind of thing that benefits from a reality check with people who cleared the review last month. If the bar has moved, I want to know before a client's submission finds out for us.

How far has MCP spread inside the ecosystem?

The last question is about plumbing: how widely is the Model Context Protocol actually being adopted inside the Salesforce ecosystem, versus talked about? MCP is a genuinely useful way to connect AI to systems under real controls, and the interesting question is not whether it is promising, but how many teams are shipping with it and what they have learned doing so.

We build Agentforce work that uses this kind of connectivity, so the practical adoption curve matters to us directly. I want to leave the week with a clearer sense of where MCP sits between hype and habit inside this specific ecosystem.

Want to compare notes in San Francisco?

If you will be at Dreamforce and any of these questions overlaps with what you are working on, I would genuinely like to talk. No pitch, just two builders comparing notes. You can book 30 minutes with us during the week, and we will find a quiet corner away from the crowd.

And a promise to close on: the week after Dreamforce, we will publish the honest takeaways, what was actually announced and what it means for CPQ and Agentforce buyers, based only on verified announcements rather than the hype cycle. This post is the questions; that one will be the answers.