Best of LinkedIn: Salesforce's Dreamforce 2025

00:00:00: This episode is provided by Thomas Allgaier and Frennis, based on the most relevant LinkedIn post talking about Salesforce Dreamforce.

00:00:08: Frennis supports ICT enterprises with market and competitive intelligence, decoding emerging technologies, customer insights, regulatory shifts, and competitor strategies.

00:00:18: So product teams and strategy leaders don't just react, but shape the future.

00:00:23: Welcome to the deep dive.

00:00:25: our mission today.

00:00:26: Well,

00:00:26: it's to give you a shortcut, really, straight to the strategic core of enterprise AI as seen at Dreamforce twenty twenty five.

00:00:32: We've looked at the key announcements, the shifts, the perspectives, specifically for tech professionals like you focused on digital transformation.

00:00:40: Yeah, and if you sift through all the noise, I mean hundreds of posts, one single concept that completely dominated everything.

00:00:46: The agentic enterprise, that's the big story.

00:00:48: It ties together AI, data workflows.

00:00:51: Everything, it's signaling a really major pivot.

00:00:53: How so?

00:00:54: Removing decisively, it seems, from AI assistance, you know, like co-pilots helping out towards AI action, systems that can actually execute complex tasks autonomously, business-critical stuff.

00:01:04: Okay, the agentic enterprise vision, let's dig into that straight away.

00:01:08: Is this actually realistic for, say, the next year?

00:01:12: I saw Lars Hokey-Bother mention the goal is this picture of... humans and AI agents working truly hand in hand.

00:01:19: It's definitely ambitious, yeah.

00:01:21: But it frames the whole architecture they announced.

00:01:23: And what's really fascinating is how the philosophy behind building AI seems to be changing to make this happen.

00:01:29: How do you mean?

00:01:30: Well, Pascal Bournet, the AI expert, made a strong case that the future runs on context, not prompts.

00:01:36: It's not about writing the perfect prompt anymore.

00:01:38: It's about engineering the perfect context for the AI to understand and act within.

00:01:43: Okay, that's a huge distinction for anyone actually building these systems.

00:01:46: If context is the new currency, what does that mean practically for like enterprise data strategy.

00:01:54: It means fragmented data is basically a death sentence for an agent.

00:01:57: Pascal Barnet shared a really striking example.

00:02:00: Oh yeah.

00:02:01: Yeah, and AI failed simply because the information it needed was scattered.

00:02:06: It was in a table, a chart, and a paragraph, all in the same PDF.

00:02:11: The data points were there, but the unified engineered context of the business reality was missing.

00:02:18: So the mandate now, and this was apparently echoed by the Dell team too, is engineer the context.

00:02:22: And what's driving this?

00:02:24: Yeah.

00:02:24: This urgency, it sounds pretty intense, almost like an internal push.

00:02:28: It

00:02:28: really is.

00:02:29: Both Helen Yu and Pascal Burnett pointed towards Michael Dell, the CEO of Dell Technologies.

00:02:34: He apparently challenged his team, saying something like, a new competitor is coming, and the only way we prevent that is we become that company ourselves.

00:02:41: Okay, yeah, that kind of top-down imperative definitely forces change.

00:02:45: And exactly, it forces immediate architectural shifts.

00:02:48: Despite that pressure from the top, we have to face the reality check on adoption, right?

00:02:52: Mark Benioff himself apparently observed technology innovation is outpacing customer adoption.

00:02:59: Enterprises just aren't keeping up with the tech itself.

00:03:01: And you see that hard truth reflected in the current reliability numbers.

00:03:06: Joshua B. Lee cited a figure that, well, it should make any leader pause.

00:03:10: Agents currently have a seventy percent failure rate.

00:03:16: That huge reliability gap really underlines why things like governance, trust, visibility into why an agent does what it does.

00:03:25: They're not just nice to have.

00:03:26: They're absolutely mandatory, foundational.

00:03:28: Right.

00:03:29: So given that need for trust and accountability, let's put it to the platform shifts.

00:03:33: The architecture designed to deliver that context.

00:03:37: Salesforce rolled out Agent Force Three Sixty.

00:03:39: That's positioned as the unified layer, yeah.

00:03:41: to operationalize these enterprise AI agents.

00:03:43: Exactly.

00:03:44: It's the framework built with the agentic capabilities threaded through every product, designed, as Sebastian Viennial stressed, to connect the human and the AI agent within one trusted system.

00:03:54: Okay.

00:03:54: But that framework is totally dependent on a solid foundation.

00:03:58: Which brings us to that crucial architectural piece the rebranding of data cloud

00:04:02: to data three sixty.

00:04:03: What's the real technical significance there for our audience?

00:04:06: It's got to be more than just a name change.

00:04:08: Oh, definitely.

00:04:09: data.

00:04:09: three sixty is the technical engine designed to unify both structured and unstructured data.

00:04:14: That unification is what provides the rich intelligent context that agents need to actually move beyond simple tasks.

00:04:22: So it tackles that fragmentation problem you mentioned earlier

00:04:25: precisely.

00:04:25: Rebecca Respby and Mark Pastors really emphasize that this is the necessary fuel for agent accuracy.

00:04:31: It stops the fragmentation at the source.

00:04:34: And it sounds like it's already showing results.

00:04:36: Averil Jain shared the booking.com example using data.

00:04:39: Three sixty to unify partner data, which led to smarter B to B segmentation and growth.

00:04:45: Yeah, real world value.

00:04:46: And then there's the collaboration layer.

00:04:48: Slack got a major reframe.

00:04:49: Let's go.

00:04:50: It's now being positioned pretty unequivocally as the agentic OS.

00:04:55: Interesting.

00:04:56: Jerry Fagel noted that Slack is becoming the central hub that orchestrates these autonomous agents.

00:05:00: It brings people, data apps, and the AI agents together right into the conversation, kind of the conversational command center, you know?

00:05:07: Makes sense.

00:05:08: Now, something for the developers, and this sounded like a really big deal, the launch of that new deployment mode, hashtag run relevant tests, that addresses a long standing pain point.

00:05:18: Oh,

00:05:19: a decade long one.

00:05:20: It truly does.

00:05:21: Nicholas Villamy highlighted this.

00:05:23: Enterprises with huge huge code bases faced this awful choice.

00:05:27: Run all the tests and wait hours, maybe days.

00:05:31: Or run just a subset and risk breaking production.

00:05:34: This new mode is smart.

00:05:36: analyzes dependencies, detects, and runs only the specific test classes that are actually impacted by a change.

00:05:41: So faster deployments, but still safe.

00:05:43: Exactly.

00:05:44: Massive step forward for velocity and safety.

00:05:46: And

00:05:46: helping with that development velocity is Agent Force Vibes, the free AI coding agent.

00:05:52: Charles W confirmed it's a VS code extension.

00:05:54: Yeah, and this is moving beyond just standard code completion, the co-pilots we know, they're calling it vibe coding.

00:06:00: The idea is teams can build whole apps, flows, data models, just by describing what they need.

00:06:05: So more declarative AI driven development.

00:06:07: Right.

00:06:08: Shifting the focus and all these developer capabilities are sort of being brought together under the Apex.

00:06:14: Okay.

00:06:14: Which takes us nicely into theme three.

00:06:16: The new products actually hitting the market and their use cases.

00:06:19: Yeah.

00:06:20: Agent Force IT service was a key launch, wasn't it?

00:06:22: Champion by Moodoo Sudhakar and Jonathan Parsons.

00:06:25: Yes.

00:06:26: And the strategic insight here, I think, is that Salesforce is prioritizing internal efficiency first.

00:06:33: The easiest place to show value with agentic AI isn't necessarily the super complex external customer journey.

00:06:40: Right.

00:06:40: It's immediate, measurable internal cost savings.

00:06:43: through automated IP support.

00:06:45: That's the low hanging fruit.

00:06:46: So

00:06:46: this IT service solution, it blends human and AI agents.

00:06:51: Handles requests right in Slack.

00:06:53: Correct.

00:06:54: The AI agent handles the routine stuff automatically, but crucially when it needs to escalate a complex issue to a human.

00:07:00: It passes along the context.

00:07:02: Exactly.

00:07:03: The human agent gets the full picture.

00:07:04: All the context already gathered.

00:07:06: Helen, you mentioned FedEx is already using these IT agents to resolve employee issues instantly.

00:07:11: Impressive.

00:07:11: Beyond general IT, though, there was a focus on verticals, too.

00:07:14: Specifically, Agent Force Life Sciences.

00:07:16: Yeah, that evolved out of the old LS engagement cloud.

00:07:19: And the key thing there is the focus on regulated workflows.

00:07:22: That's

00:07:22: right.

00:07:23: David Derriere noted it provides fully built-in workflows designed for regulated industries.

00:07:29: And Samar Ansari from Takeda had a really strong message for the industry.

00:07:33: What was that?

00:07:34: Basically challenging leaders to put words into action, to stop using we're regulated as an excuse for slow adoption and really embrace a patient first mindset with these new tools.

00:07:45: That's a powerful call to action.

00:07:47: And finally, in terms of products, commerce.

00:07:49: Big moves there too.

00:07:51: Caleb Bryant confirmed Salesforce Commerce Cloud is now positioned as the world's first natively agentic commerce platform.

00:07:58: Okay,

00:07:58: what does that mean in practice?

00:08:00: Well, agent force personal shopper and merchant agents are now generally available.

00:08:04: Agent force search agent is coming.

00:08:06: But the really strategic thing to watch is the agentic commerce protocol.

00:08:09: A protocol.

00:08:10: Yeah, it's a partnership with open AI and Stripe.

00:08:13: It seems designed to set the stage for truly integrated autonomous commerce experiences that span different platforms.

00:08:20: Interesting.

00:08:20: Okay.

00:08:20: So all this speed, all these new agentic capabilities, it naturally leads to theme four, the ecosystem governance.

00:08:29: and well, the human side of things.

00:08:31: Absolutely.

00:08:32: Robust governance is non-negotiable with this level of automation.

00:08:35: And

00:08:35: partnerships are key to the ecosystem strategy, it seems.

00:08:39: The open AI integration got a lot of attention.

00:08:41: Laura, Sophie Ballther and Jujval Shah confirmed that GPT-V powered features.

00:08:46: Yes,

00:08:46: and allowing Salesforce CRM data to be accessed right inside ChatGPT, that's a massive leap in data connectivity, changes where and how that data can be used.

00:08:55: So they're still using Anthropics cloud models too, right, for regulated industries.

00:08:59: Correct, it's a multi-model strategy.

00:09:01: But trust needs transparency, especially with that, that seventy percent failure rate we mentioned lurking in the background.

00:09:06: Right.

00:09:07: Which is where Agent Force Observability comes in.

00:09:09: Ujval Shah and Julie Marques highlighted this.

00:09:11: Think of it as the accountability layer for the CIS admin.

00:09:14: So it gives admins visibility into what exactly?

00:09:18: Agent

00:09:18: performance, confidence scores, and crucially, the reasoning trails.

00:09:23: The audit log of why an agent took a specific action.

00:09:27: Hitesh Sahidwala really hammered this point home.

00:09:31: You simply cannot scale this tech responsibly without that level of demonstrable compliance and trust.

00:09:38: Makes total sense.

00:09:39: But then there's the human ecosystem itself.

00:09:41: There seemed to be some maybe tension.

00:09:43: David Giller noted a feeling that Salesforce admins weren't getting the same recognition in the keynote.

00:09:48: Yeah, with all the agent force this, agent force that, focus the question kind of hung in the air.

00:09:53: The question being?

00:09:54: Well,

00:09:54: who cleans up the mess?

00:09:55: Who detangles things when, you know, maybe poorly governed agents or users without enough training start deploying these powerful, agentic flows everywhere.

00:10:04: Ah, right.

00:10:05: The admin's job fundamentally changes.

00:10:07: It's less about managing records and more about managing autonomy, potential payoffs even.

00:10:12: That's a real anxiety.

00:10:13: It is, but, you know, the spirit of the Salesforce community also provided a really nice counterpoint.

00:10:17: How so?

00:10:18: Sebastian Jimenez shared that great story about Rilla.

00:10:20: Remember the sudden rainstorm?

00:10:22: Oh yeah.

00:10:23: They handed out umbrellas.

00:10:24: A thousand umbrellas, just like that.

00:10:26: A fast, scrappy, totally human act of customer obsession.

00:10:29: That's

00:10:29: fantastic.

00:10:31: And it stands in such sharp contrast to all talk about machine autonomy, doesn't

00:10:34: it?

00:10:35: It really does.

00:10:36: A tangible reminder that even as AI runs the architecture, that human connection, that customer focus still drives the actual experience.

00:10:45: A perfect note to end on if you enjoyed this deep dive, new episodes drop every two weeks.

00:10:49: Also check out our other editions on cloud insights, sustainability in green ICT, digital products and services, health tech, defense tech, ICT and tech insights and artificial intelligence.

00:10:59: So

00:11:00: Dreamforce, twenty twenty five left no doubt.

00:11:03: AI is now, well, it's the foundation of modern business.

00:11:05: We've seen Agent Force, Three Sixty, Data, Three Sixty, run relevant tests.

00:11:10: The capability is accelerating at an incredible pace.

00:11:13: But the real question, the one everyone leading digital transformation needs to grapple with, and Javier Ramirez put it perfectly, can we actually scale AI agents responsibly, you know, integrate that governance and trust at scale?

00:11:24: Or are we still kind of stuck just showing off impressive but isolated cool demos?

00:11:29: That really is the ultimate test of the agentic enterprise vision.

00:11:33: Thanks for diving deep with us today.

00:11:35: Be sure to subscribe so you don't miss our next exploration.