The reality behind Salesforce’s agentic marketing vision.

I made it to Chicago for Salesforce Connections 2026. After catching up with my fellow Salesforce Marketing Champions, we piled into the convention center for the main keynote, eager to see exactly where the platform is heading next.
The vision.
The core message of Connections 2026 was unmistakable: The agentic era is officially here.
Salesforce sold a beautiful vision of the future centered around “marketing makers.” A world where autonomous AI agents do the heavy lifting while we, the marketers, stay firmly in control.
Using low-code or no-code interfaces, plain-text descriptions, and tools like Slack, the promise is that anyone can spin up pipeline generation and customer acquisition strategies out of thin air.
And honestly? The live demos were impressive. We watched an artificial intelligence (AI) Sales Development Representative (SDR) qualify a lead over live video, smoothly handle objections, and book a meeting.
We saw agents stitch together fragmented Customer Relationship Management (CRM) data, a six-month-old spreadsheet, and scattered Slack threads into a polished Quarterly Business Review (QBR) summary, complete with a Tableau visual, in just a few minutes.
We even watched a marketer stand up a highly segmented, on-brand campaign with built-in guardrails designed to auto-pause if the click-through rate fell below a certain threshold. It was seamless, slick, and undeniably cool (with Anthropic’s Claude doing some heavy lifting underneath, for what it’s worth).
The reality check.
If you are already deeply embedded in the Salesforce ecosystem, this might be your moment to switch it on and at least try agentic marketing out.
But here’s the thing: I’ve been building in this platform since 2012. As a Salesforce Marketing Champion who spends my days deep in the trenches of Salesforce Marketing Cloud (SFMC), Data Cloud, and Pardot with real clients and real data, I know that the keynote stage and the actual implementation backlog look like two very different places.
Glossy demos are fantastic, but they notoriously leave out the fine print. As I head home from Chicago, three major questions are on my mind:
- How many SKUs, licenses, credits, and seats does it actually take to assemble the workflow I just watched on stage?
- How pristine does your data have to be before any of this works? Every demo assumes a perfectly governed semantic layer. Whose everyday reality is that?
- How long does it take to go live? Can a client team realistically handle this in-house, or do they need to partner with a Solution Implementation expert (like our team at BarkleyOKRP) to make it happen?
No matter whose AI you buy, your data is the ultimate prerequisite.
The flashy front-end isn’t the magic; the magic is what’s underneath. Your data, your architecture, and your guardrails grounded in your own platform.
If your data foundation isn’t ready, the agents won’t be either.
My challenge to Salesforce.
During the keynote, Salesforce called marketers “control freaks.” Honestly? That’s totally fair. We are control freaks. But if you want to win over a room full of us, give us the one thing that actually builds trust: the truth about what it takes to execute.
So here is my challenge to Salesforce: Follow every keynote demo with the receipts. Give us the exact platform Stock Keeping Units (SKUs), the required seats, and a real-world implementation playbook to replicate what we see on stage.
Show us that, and I’ll get incredibly excited.
Until then, count me impressed…but not quite sold. If you were at Connections and you’ve already cracked the platform-and-licensing math on this, hit me up. I’d genuinely love to see those receipts.
__
Ready to turn flashy AI demos into a real-world roadmap for your business? Let’s talk about how BarkleyOKRP can help you audit your data layer, navigate the licensing math, and build a scalable CRM strategy for this next era of marketing. Reach out to Pat LaCroix, EVP, Media + Growth, at placroix@missiononemedia.com.