Agentforce 2.0: AI agents of the next generation

The presentation of Agentforce at Salesforce Dreamforce in September 2024 caused quite a stir. Now Salesforce has given the platform a major upgrade: Agentforce 2.0 expands reasoning, can be integrated even more seamlessly into systems and adapts dynamically. This allows companies to use AI agents in an even more intelligent and versatile way.

Five new Agentforce 2.0 features take center stage: a larger library of pre-built functions, cross-system workflows, integration with Slack, a more powerful engine for deeper analysis and the ability to create agents in natural language.

“Agentforce 2.0 takes our revolutionary Salesforce digital labor platform to another level, with new reasoning, integration and customization features that supercharge autonomous agents with unprecedented levels of intelligence, precision and accuracy.”

Marc Benioff, CEO Salesforce

Before we get into the new Agentforce 2.0 features, let's take a quick look at the core functions:

Agentforce is a platform that enables companies to quickly and easily integrate artificial intelligence into all business processes. And without in-depth programming knowledge.

Autonomous, intelligent AI assistants, known as agents, support employees, for example in sales, customer service and marketing. They take on routine tasks, automate workflows and carry out research. They analyze data, make data-based decisions and thus give employees more capacity for complex tasks and personal customer care.

You can read more details about this in our article about Salesforce Agentforce.

The library of ready-made agent functions, known as skills, has grown significantly with Agentforce 2.0. In addition to the CRM system, Salesforce's core product, the new functions also cover Slack, Tableau and the AppExchange ecosystem. Companies can now implement a wide range of automations more quickly and no longer have to invest a lot of time in developing individual solutions.

External partners can also provide their own skills via AppExchange. This further expands the collection of ready-made functions.

Agentforce 2.0 facilitates cross-system workflows. A key role is played by MuleSoft for Flow: Companies can use the low-code platform to seamlessly connect workflows across different systems.

A new feature is the MuleSoft API Catalog, in which all available interfaces are compiled centrally. This allows developers and administrators to find relevant APIs more quickly and manage them more easily.

Another Agentforce 2.0 feature is the MuleSoft Topic Center. It embeds additional metadata in the APIs to automatically provide Agentforce with the correct and relevant information. Salesforce is thus consolidating its agent-first strategy, increasing interoperability and accelerating the implementation of complex integration projects.

Agentforce goes Slack! With the next generation of AI agents, smart assistants can be brought into any conversation for the first time. Users can launch the agents from the Agentforce Hub or address them via mentions and direct messages in Slack.

With Slack Enterprise Search, agents can access public and authorized information in Slack and provide well-founded answers. Anyone who also has standardized queries, such as status information on ongoing projects or wants to initiate a data query, can use ready-made Slack skills.

New Tableau skills offer optimized data visualizations and predictions so that users can better understand the agents‘ responses.

The Atlas Reasoning Engine, Agentforce 2.0‘s neural network, now processes queries even more thoroughly. In addition to simple status queries, the AI agents can also support complex decision-making processes – for example, which technology solution the agent recommends, taking into account the current company situation and the overall economic situation.

These complex queries are made possible by the so-called agentic loop. This means that the AI agents check and improve their answer iteratively until the most precise result is available. This is done automatically and requires no manual intervention.

The high-performance Atlas Reasoning Engine is supported by an extended Retrieval Augmented Generation (RAG). From now on, even larger volumes of unstructured data can be searched using additional metadata. This increases both the accuracy and relevance of the search results.

With Agentforce 2.0, it is now even easier for all teams to create their own agents. In the Agent Builder, all you need to do is enter the instructions in normal language. The tool automatically suggests suitable skills and actions from the connected library.

This significantly shortens the development of AI agents and avoids recurring rounds of coordination with IT managers.

Once the AI agents have been set up, they can be integrated into existing tools and workflows without any further effort or training.

Salesforce is currently driving Agentforce forward with a strong focus on connecting people and AI - with autonomous, intelligent AI assistants for every employee.

Agentforce 2.0 is far more than just an update. The platform opens up new opportunities for companies to integrate AI agents into their processes at an even lower threshold. At the same time, it increases the level of automation enormously. Complex workflows can be easily mapped, results are delivered precisely and agents can be created intuitively – in line with the dynamic requirements of modern organizations.

Companies that want to make their digital processes more efficient, exploit the potential of their Salesforce CRM system and stay one step ahead of their competitors cannot ignore Agentforce 2.0.