Salesforce Dreamforce ended on September 19th and was packed with various announcements and highlights, most notably the premiere of AgentForce and what it means for the future development of the Salesforce platform.
1st keynote speech by Marc Benioff
Marc Benioff’s keynote sets the scene for the Salesforce Dreamforce. It’s amazing what Salesforce has put together again.
If you want to watch the keynote again, you can find it here on YouTube or on Salesforce+. But now to the highlights of Dreamforce…
2. agentforce
The biggest highlight was definitely the Agentforce and how organizations will be able to develop independent agents in the future and how these will interact with the core platform. Agents such as service, sales or marketing agents are provided for this purpose, which bring the entire AI power with them and interact with users or customers. This was also demonstrated in the keynote. I found the interaction with Slack particularly exciting.
In my opinion, the following announcements about Agentforce are worthy of note:
- The heart of Agentforce is called Atlas. It is a so-called reasoning engine that simulates human thought and action. From task entry to execution, this engine checks every step for correctness and relevance. In this way, Agentforce can make precise and objective decisions and complete tasks independently.
- The Agent Builder is used to create agents that are used for automation, API queries or development on the Salesforce platform. Together with Prompt Builder and Model Builder, Agent Builder becomes Agentforce Studio.
To better understand Agentforce and how collaboration on the platform works, it is helpful to take a look at the keynote. Among other things, a service agent is presented there and how it is customized via click-no-code. This is demonstrated in a very appealing way (Salesforce-like) and gives a very good impression of what is possible and how Agentforce differs from assistance systems.
However, it was also made clear that Atlas is based on data cloud data, which seems enormously important because, after all, AI is only as good as the data on which it is based. This is why the data cloud continues to move into focus.
3. data cloud
The Salesforce Data Cloud has been in the spotlight at Salesforce in recent years. Initially as a CDP for marketing use cases. The name was then changed (does anyone remember “genius”) and now the Data Cloud is the data foundation that underpins every Salesforce “cloud” product. Continuous development and improvement. It is also emphasized that the Data Cloud is a system of reference, i.e. it does not enforce its own data storage, but only accesses existing data via connectors.
Even if the announcements seem trivial, they are all the more important for the basis of AI because they form the foundation:
- The processing of other unstructured data formats is possible: Native processing of video and audio formats is possible. Since 90% of company data is unstructured (as opposed to structured in data records, objects and object relationships), companies miss out on a wealth of information that cannot be captured in conventional data structures. With this announcement, video and audio (e.g. demos, customer service calls, voicemails) can be analyzed and insights gained.
- More connectors: With a further 50 prefabricated connectors, the total number now exceeds 200 prefabricated connections.
- Real-time capabilities in fractions of a second: The Data Cloud will be able to work incredibly fast in all the processes it handles – from data ingestion to activation, all in fractions of a second. In my view, this will give the data cloud another real boost!
The interaction between Data Cloud and Agentforce is summarized again below.
4. slack innovations
As already mentioned above, I personally think that one of the highlights is the further development of Slack and the integration into Agentforce.
Agents are treated as equal members of a channel and also provide input to discussions that take place in a channel. In my view, this is even more special because structured data as well as unstructured data is used. In the example below, 25 people and 3 agents are active in the channel.
Each agent has their own profile with a list of skills. But depending on the role, the task profile is also described (what an agent can/should do) and what not (so-called guardrails).
- Third-party AI agents: Agents and assistants from partners like Adobe, Anthropic, Cohere, Perplexity and others can be deployed in the same trusted, secure environment in Slack.
- Salesforce Channels: What is finally coming is the integration of CRM data within the channels. This means that an approval can be initiated directly from the communication, an order history can be viewed or a support ticket can be opened. Below you can see a use case of a merchandising team where an order has been placed.
All in all, a really good leap for internal communication and internal processes, especially as the agent reacts without a direct prompt and makes suggestions within the channel that can then be discussed.
5 Salesforce Foundations
With Salesforce Foundations, Salesforce is launching a free feature set that includes new Sales, Service, Marketing, Commerce and Data Cloud features, as well as an update to the user interface. Salesforce hopes that this will give companies a taste for the new products and that they will be able and willing to use the power of Agentforce.
All details about the free license can be found here in the Salesforce Newsroom article (as some features have limits of course),
6. out-of-the-box agents for industries
Salesforce recognized early on that the use of CRM can be increased if the needs of individual industries are already taken into account when setting up the CRM. This competitive advantage will also be further expanded with Agentforce. Over 100 industry-specific prompts are provided for Agentforce, which can be used.
Simply scan the QR code to access the Industry AI Use Case Library.
7th Agentforce World Tour
Salesforce will market Agentforce worldwide until the end of the year with the World Tour, which begins in October. A visit is definitely worthwhile. I am personally looking forward to Agentforce on November 6 in Düsseldorf.
Summary
Marc Benioff mentioned it several times in his keynote that this is probably the most important Dreamforce of all time. With Agentforce, Salesforce has also created another milestone here.
It remains to be seen how this will now be put into practice, but I am confident that more agents will support us in our daily work in the future!
The announcement for the next Dreamforce 2025 is still pending, I’m curious to see when and where it will take place!
