Dreamforce Predictions, Pragmatic AI, and the Rise of Decision Intelligence | CRTV Episode 138

This week's episode of ConstellationTV covers a lot of ground: a Big Debate on what to expect from Salesforce at Dreamforce, a ShortList spotlight on B2C marketing automation, a deep dive into Oracle's database and agentic AI strategy ahead of Oracle AI World, and a look at the fast-growing decision intelligence platforms market.

Big Debate: What's Coming at Dreamforce

Two weeks out from Dreamforce, analysts Michael Ni, Holger Mueller and Esteban Kolsky debated where Salesforce stands heading into the conference. Ni argued that Salesforce enters this year's event in a stronger position than last year, noting that the SaaS thesis that AI would fully disintermediate platforms hasn't played out as some predicted. But he flagged a real risk: if Salesforce loses visibility into user clicks and voice interactions to AI intermediaries like Anthropic, it could lose ground in monetization.

Kolsky pushed back on the idea that Salesforce's self-service data advantage is enough to carry the platform. His take: Salesforce needs to double down on governance, execution, and agent management rather than trying to outcompete pure-play data platforms. He went further, predicting that Agentforce itself could be retired within two to three years if it doesn't deliver.

The debate ended without full agreement, but that's the format. Tune in over the next two weeks to see how the actual Dreamforce announcements measure up.

ShortList Spotlight: B2C Marketing Automation

Analyst Liz Miller broke down what's changed in B2C marketing automation over the past two years. Her take: the category has moved well past list size, segmentation, and channel orchestration. Those capabilities are now table stakes. What separates ShortList-caliber vendors today is the ability to deliver connected customer experiences that drive durable, profitable relationships.

Miller spotlighted Pegasystems' Customer Decision Hub as a standout on this year's list, centering her analysis on the word "pragmatic." Pega's approach applies generative AI at design time, where deep reasoning justifies the cost, while keeping repeatable, high-volume decisions deterministic and automated rather than probabilistic. The result, in Miller's framing, is a platform built at the intersection of what's right for the business and what's right for the customer, rather than optimizing for one at the expense of the other.

Oracle AI World Preview: Database Meets Agentic AI

Ahead of Oracle AI World, Constellation analysts Holger Mueller and R "Ray" Wang unpacked Oracle's recent wave of database and agentic AI announcements. The headline architectural decision: Oracle keeps transactional data inside the Oracle database rather than migrating it into a separate data lakehouse. Unstructured content is ingested into an Oracle AI Data Lakehouse, is vectorized by the Oracle AI Vector Database, and the resulting vectors are then written back to the transactional database. CIOs and CISOs favor this model because it avoids the need to recreate data security and governance profiles in a second system.

Other announcements covered included the AI Database Private Agent Factory for building agents directly on Oracle infrastructure, deterministic trusted answer search to eliminate the risk of hallucinations, voice-driven interactive reports in APEX that stay grounded in existing data, and MCP server support for both on-premises and cloud (autonomous AI database) customers.

The bottom line, as the analysts summed it up: Oracle's differentiation is that customers don't have to move or rearchitect their data to get agentic AI capabilities, while most competitors require exactly that.

ShortList Spotlight: Decision Intelligence Platforms

Ray Wang closed the episode with Constellation's Decision Intelligence Platforms ShortList, a market he projects will reach $42.3 billion by 2031. These platforms run mission-critical, self-driving, self-healing business processes, spanning use cases like campaign-to-lead, procure-to-pay, and hire-to-retire, and trace their lineage back to RPA, process mining, BPM, and journey orchestration.

Wang spotlighted Celonis as one of the vendors on this year's list, highlighting its May 2026 acquisition of Ikigai Labs and the launch of the Celonis Context Model (CCM). The acquisition brings Ikigai's patented Large Graphical Model technology, MIT-backed research focused on tabular and time-series data, into Celonis' platform, adding real-time digital twin capabilities, scenario simulation, and dynamic planning on top of Celonis' existing process intelligence graph. Wang also noted the architecture's zero-copy integrations with AWS, Databricks, Microsoft Fabric, and Snowflake, delivering significant query performance gains.

As agentic workflows become more embedded in enterprise operations, Wang frames decision intelligence platforms as increasingly central to how businesses actually run, not a peripheral analytics layer.