Salesforce announced the launch of Agentforce Atlas, the latest version of its Agentic AI solution, in time for Dreamforce. Marc Benioff, Salesforce’s Chairman and CEO, believes that with this release, Salesforce is delivering the third wave of AI.
While this is a significant announcement, Benioff believes there is more to come in the future. He said, “Yes, we’re very excited at the fourth wave. We can start to see it: the wave of droids that’s coming. Maybe this mythical fifth wave of AGI will come one day as well.”
Unlike some previous announcements, this one is actually here and available at Dreamforce. Salesforce revealed several live customers, including Wiley and Saks, and demonstrated how simple it is to build new skills onto the existing implementation with the Saks installations.
Benioff also has an ambition to have even more customers adopt Agentforce at Dreamforce. He said, “One of the cool things is we are going to do the largest deployment in the history of our industry, of agents at Dreamforce.
“We will turn on as many customers as we can, and our goal is at least 1000. So that when they leave Dreamforce, we will have 1000 working companies all running these next-generation agents and have 1000 agent forces.”
Agentforce Atlas is the second iteration of Agentforce. In the summer, it released autonomous agents for Sales Cloud and Service Cloud, which it has now named Agentforce. This new release will make Agentforce available across all fifteen of Salesforce’s clouds.
It is also better. One of the early adopters is the healthcare organisation Kaiser. Benioff revealed, “Kaiser on Atlas is seeing more than 90% of these patient inquiries resolve their CTO said to me, it’s remarkable. It’s really what we have always wanted AI to be. We’re going to give those doctors their time back.”
What is Agentforce
Salesforce has put a lot of thought into creating Agentforcce, its Agentic AI solution for business. It has the guardrails and data platform in place. Next week, it plans to release details of Salesforce Foundation, a new version of Customer 360 on which all its recent acquisitions are now built. Coupled with the new release of Data Cloud and the AI platform, it now has a platform ready to create Agentic AI.
Clara Shih, CEO of Salesforce AI, explained what Agentic AI can do, “We’re going from content generation and analysis to automating actions. In the generative phase, you might ask a copilot to write an email for you to a customer.
“In the agentic phase, you can ask a harder question: ‘What should I do with all of my customers?’ Maybe it’s email, maybe it’s picking up the phone and calling, maybe it’s sending a text message. That’s really what agents can do: They can take a higher order question, break it down into a series of steps, and then execute each of those steps.”
Agentforce is able to interrogate the Data Cloud, which can include Structured and Unstructured Data. With its Zero Copy capability, it can access data from other sources, such as Snowflake, DataBricks, BigQuery, or Redshift.
The Atlas Reasoning engine then retrieves the data, plans, evaluates, and refines it in the context of the actions it can initiate. Salesforce customers can define actions within Salesforce Flow, Mulesoft, or Apex across any channel.
There was an impressive demonstration of the Saks Agentforce. Patrick Stokes, EVP of Product and Industries Marketing at Salesforce, demonstrated how easy it was to adapt Agentforce to deal with new use cases that might have needed escalation to a human. The Agentforce Agent, Sophie, interacts using voice in a clear, humanlike way, having first identified itself as AI.
It was like talking to a human agent, but one that had full corporate knowledge without the restriction have having to switch screens or look through historical transactions.
With no code, he was able to add a new delivery option of collect from the store, using the API integration with Mulesoft and predefined Flow templates by simply switching them on.
Agentforce agents is not only available from Salesforce, several partners have been working on the new Agentic AI as well. Expect announcements from these partners soon, who include Amazon Web Services, Box, Certinia, Copado, Coupa, Google, Honeywell, IBM, and Workday.
How big can AgentForce become, and is there a catch
Salesforce has the ambition to have a billion agents worldwide. They are already well on the way. Initial customers include Disney, Wiley, OpenTable and ADP. ADP alone has 1.5 million customers, serving 42 million consumers, and OpenTable has 160 million consumers.
Benioff revealed that Agentforce is available at $2 per conversation as a starting point. It will be interesting to see how that pricing scales as the conversations grow. Wiley is already impressed, though.
The first iteration of Agentforce has helped Wiley see a 40-50% increase in case resolution, outperforming their old bot. There are large cost savings as it normally has to hire a larger number of temporary workers to cover the peak period.
Kevin Quigley, Senior Manage and Agentblazer at Wiley, commented, “We’re able to cover virtually everything in our knowledge base because it has that reasoning, and it can do that search and apply the case-by-case responses. We’ve seen an over 40% increase in our case resolution when you compare the agent to our old chatbot.”
While Agentforce looks set to revolutionise service and many other functions, there is a catch. It is as efficient as it has proven at Wiley there are at least two concerns.
The first is that as Agentforce is trained, fewer people will understand how decisions are made, leading to a loss of corporate intelligence at the ground level. While Salesforce demonstrated that it offers explainable AI so business users can understand the logic of the decisions, will future workers understand that logic?
The second is, as Wiley showed, less hiring requirement. What will this mean for the economy? There will be fewer jobs and potentially fewer entry-level positions.
In the long term, this might not be an issue, but in the short term, as was seen with the first phase of the Industrial Revolution, the quality of life and work available fell sharply. Jobs on the farms evaporated, well before the industrialisation of the cities, leading to significant poverty.
Enterprise Times asked Salesforce about this during the briefing and is still awaiting a response. The article will be updated when a response is received.
Enterprise Times: What does this mean
Agentforce is a game changer. Many firms will have substantial, possibly seasonal, requirements where Agentforce can replace a significant number of hires. The price point will be attractive for many.
However, customers will need to consider the human handoff where needed. While Agentforce can deal with queries 24×7, which is a good thing for customers, will it mean they need to hire shift workers to deal with issues over the same period?
There will be more about Agentforce at Dreamforce, and it will be exciting to see what Salesforce has planned for the future as well.
This third phase of AI could become incredibly disruptive, though. While it will benefit organisations, it will be interesting to see the impact on workers and the workplace in a few years.
