Agentforce 2.0 has finally been revealed with new reasoning, integration, and customization features to “supercharge” Salesforce’s flagship AI.
At a long-awaited unveiling in San Francisco on December 17, the cloud giant unveiled the next iteration of its agentic product, introducing a new library of pre-built agent skills spanning CRM, Slack, Tableau, and partner-developed skills on the AppExchange.
What Is Agentforce 2.0?
The new release introduces a new set of pre-built skills and workflow integrations for rapid customization, the ability to deploy Agentforce in Slack, and advancements in agentic reasoning and retrieval augmented generation (RAG).
Chair and CEO of Salesforce, Marc Benioff, said in a statement ahead of its launch:
“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.
“The demand for Agentforce has been amazing – no other company comes close to offering this complete AI solution for enterprises. We’re seamlessly bringing together AI, data, apps, and automation with humans to reshape how work gets done. Agentforce 2.0 cements our position as the leader in digital labor solutions, allowing any company to build a limitless workforce that can truly transform their business.”
Some key takeaways and dates from the unveiling of Agentforce 2.0 today include:
- Agentforce 2.0’s full release will be generally available in February 2025
- Specific features for 2.0 are being released in advance – beginning today
- Skills for Tableau will be generally available from December 18
- Agentforce in Slack and other Slack functions will be generally available in January 2025
- Natural language creation of agents in Agent Builder will be available in January 2025
- MuleSoft for Flow, MuleSoft API Catalog, and Topic Center will be available in February 2025
- Enhanced reasoning and RAG will be generally available in February 2025
Salesforce name-drops companies including Accenture, The Adecco Group, Finnair, Heathrow Airport, IBM, Indeed, Saks Global, and SharkNinja when describing how businesses are “embracing Agentforce” to augment their teams with “digital labor”.
Now, Salesforce is introducing a new library of skills to bring Agentforce into “every team and workflow”, helping avoid time and integration challenges.
Agentforce 2.0 introduces a range of pre-built agent skills which span CRM, Slack, Tableau, and partner-developed skills on the AppExchange.
So what exactly are these new features that make Agentforce 2.0 so distinct from the original?
New CRM Skills
According to Salesforce, new skills for sales teams like Sales Development and Sales Coaching allow for the creation of autonomous AI agents that can nurture leads based on your particular rules of engagement. Agents are also able to join prospecting calls and provide instant feedback on interactions with customers.
Other skills include Marketing Campaign and Commerce Merchant skills, scheduling skills for service engagements, and new skills for field service workers.
MuleSoft: Take Action Across Any App or Workflow
MuleSoft now lets Agentforce work across businesses. With MuleSoft for Flow, Salesforce says it is now easier than ever to create lowcode workflows that span any system, with pre-built connectors for building multi-system workflows fast.
As teams look to turn their APIs into Agentforce actions, the new MuleSoft API Catalog lets builders and Salesforce Admins view, discover, and manage APIs across Salesforce, MuleSoft, Heroku, and any external services from one central location for rapid reuse.
The new MuleSoft Topic Center enhances this by letting teams infuse Agentforce metadata into every API they build, ensuring that any point of connectivity can be automatically turned into an Agentforce skill or action, making every API “agent-first” by default.
Tableau Skills for Analytics and Insights
New Tableau Topics and Actions deliver data visualizations and predictions for better understanding of agent responses and accurate, context-rich answers using Tableau Semantics. This unlocks new conversational analytics use cases, further lowering the barrier to accessing data.
Partner Skills Through AppExchange
Agentforce is backed by what Salesforce is calling the “first-ever enterprise ecosystem” of agent skills.
The cloud giant says this lets customers extend Agentforce with custom Topics and Actions ranging from new agent types – such as the AI Employee Service Agent – with Workday, to new, partner-built actions from Asymbl, Docusign, and Neuron 7.
Agentforce Recommends Skills
Salesforce says you can now create new agents “in seconds” using natural language descriptions.
Agent Builder now uses Agentforce to compose new agents for the work you need done by auto-generating relevant topics and instructions, while pulling from the library of skills and actions already available to you.
This, Salesforce says, will make sure teams can go live with new digital labor quickly.
Agentforce in Slack
This is a big one. Agentforce/Slack compatibility was a big topic at Dreamforce ’24, and the integration of the two is being further enhanced with the latest upgrade.
Agentforce 2.0 lets teams bring the agentic AI into any Slack conversation, Salesforce says.
Slack users can start a conversation directly from the Agentforce Hub, or @ mention Agentforce agents through DMs or in channels, tapping into their digital labor force directly in the flow of work.
Agent Builder now features pre-built Slack Actions like “Create Canvas” or “Message Channel” which let teams quickly improve existing agents or create new ones that can engage with teams in Slack.
And now, with Slack Enterprise Search, Agentforce can draw from conversational data – enhancing the relevancy of responses and actions by drawing from public and permissioned information in Slack.
Slack Actions are also being made available in Agent Builder, meaning a team can enhance Agentforce with, for example, the ability to send a direct message to provide a summary of what’s going on with a project.
It can also update a Slack Canvas when a customer asks for changes to an ongoing project.
Enhanced Atlas Reasoning Engine
The Atlas Reasoning Engine is described as the “brain” behind Agentforce, letting it retrieve data, then reason, then act.
Agentforce 2.0 will see the introduction of “enhanced reasoning” and retrieval to handle deeply nuanced questions, powered by new capabilities in Data Cloud that fuel Agentforce with greater context, according to Salesforce.
The engine can now handle a larger variety of interactions, including those with several layers that require more in-depth thought.
Salesforce gives the following question as an example: “What is the status of my portfolio?” This question will make use of basic reasoning for rapid responses.
However, a deeper question – such as “What would be the right investment vehicle for my child’s college fund based on my current income and risk preferences?” – would make use of “enhanced reasoning”.
This will use advanced data retrievers, improving the answer by refining the query, before pulling both relevant data and context-specific metadata in Data Cloud, Salesforce says.
The Atlas Reasoning Engine will assess its own response and loop through a variety of tools and sources in what the company calls an agentic loop, letting it provide a “trusted, well-researched response” or action to more complex requests, without writing custom code.
The newest version also makes use of enhanced RAG with enriched indexing. RAG allows Agentforce to find relevant information quickly in unstructured content. To power enhanced reasoning in the Atlas Reasoning Engine, Data Cloud can now enrich RAG chunks – the snippets of retrieved data – with metadata from the Salesforce platform.
Enriching RAG indexes with metadata context helps adapt Agentforce 2.0 to business conventions without extra work, improving accuracy and relevancy for requests.
Inline citations will also point to the exact sources Agentforce pulled from to answer a given question, in a bid to improve trust in its abilities.
Final Thoughts
Earlier this month, Salesforce CEO Marc Benioff had teased “Agentforce 2.0”, telling those on the company’s Q3 earnings call they would “not believe” what they’d see on December 17. A few days later the cloud giant officially announced the ramped version of its AI product, publishing a landing page for it.
Salesforce Ben had reported on December 9 how we might have already seen hints of what was to come during Dreamforce ’24, where a demonstration showed how Slack and Agentforce could work together.
We also noted how the timing of the announcement was somewhat auspicious; talk of “greater accuracy” and an upgraded Atlas Reasoning Engine came amid news from fellow tech giant Open AI that they had launched a newer version of the ChatGPT o1 model (o1 pro mode), which came as part of a $200-a-month subscription.
We asked at the time whether there would be significant enough difference between Agentforce “1.0” and this latest version to warrant the distinction of “2.0” as a wholly new product, but the updates have now been revealed to be wide-ranging and extensive – it’s fair to say that the new moniker is a reasonable one.
Agentforce is clearly the primary focus of Salesforce now. As we move into 2025, it may not be long until we start asking what Agentforce 3.0 will look like.
But for now, those in the Salesforce ecosystem and the AI sector will have a lot to get up to speed with as Agentforce 2.0 is rolled out across businesses, and the customer success stories keep rolling in.