From Agentforce 1.0 to 2.0, in just three months, Salesforce rapidly iterated on AI agents, and the company also plans to recruit 2,000 people to sell its AI software, demonstrating Salesforce's determination to do everything in AI agents.
AI is everywhere, revolutionizing work models, and startups are rising... and actively embracing AI is Salesforce's way to survive in the AI era.
Recently, US cloud computing software giant Salesforce has shown unprecedented attention to AI agents. From Agentforce 1.0 to 2.0, in just three months, Salesforce achieved rapid iteration and major upgrades in the AI field.
Since launching the AI agent product AgentForce in September, Salesforce CEO Benioff has issued a “dead order” requiring the sales team to include software in every transaction. The company also plans to recruit 2,000 people to sell its AI software.
As reported by Information on Thursday, while reviewing the product with Salesforce executives, Benioff emphasized:
This is the most important thing for the entire company: everything is agent-first, we want to be a completely agent-first company, and every transaction we make or every product we make should put agents first.
Benioff plans to achieve the ultimate goals: AI agents replace Salesforce software, such as its sales and customer service software, and its business model that charges customers subscription fees on an employee-by-employee basis.
However, the sense of urgency of being an all-in-one AI agent stems from AI's threat to Salesforce's traditional business model. For Salesforce, they either embrace AI or go “extinct.”
The threat of the rise of startups
In fact, as early as 2016, Salesforce joined the wave of AI development, launched Einstein products, and bound AI into sales software.
Despite having a first-mover advantage, Salesforce hasn't maintained its lead, and its artificial intelligence products have lagged behind some newer startups. Salesforce can't do as many operations as a new product — especially in the area of customer service.
This led to the loss of Salesforce customers. Casper, a mattress manufacturer, purchased AI agent software from Sierra. Sierra is a startup founded by former Salesforce co-CEO Bret Taylor. Despite Salesforce's efforts to dissuade customers, Casper made this decision.
People familiar with the matter said Casper executives were initially less aggressive about using the Sierra platform and chose to use it only for customer support. But Sierra's AI agent had very good results for Casper, so the company began experimenting with using it for sales.
Zeya Yang, IVP partner at a venture capital firm that invests in enterprise software companies, said:
Startups like Attio and Day.ai are challenging Salesforce by building customer relationship management software that uses artificial intelligence to capture information not included in Salesforce.
Even Salesforce's partnership with OpenAI didn't give it an advantage.
In early 2023, shortly after OpenAI released ChatGPT, Salesforce announced a partnership with OpenAI to combine the AI company's technology with Einstein. However, this partnership did not protect Salesforce from the threat posed by AI companies, and some companies are already testing LLM from OpenAI and others to see if it can replace Salesforce software.
All in AI agents
Recently, Salesforce began making full use of AI.
In September, Salesforce released a new artificial intelligence agent, Agentforce (a software tool that can automate business tasks). In December, Agentforce 2.0 brought a number of breakthrough features, and its core improvement was its enhanced inference engine.
In just three months from Agentforce 1.0 to 2.0, Salesforce demonstrated rapid iterations and significant upgrades in the AI field. The new version has better reasoning skills, which can help it provide more detailed answers to investment or recruitment questions.
In terms of product sales, Benioff said in a recent interview:
It's been less than two months since Salesforce began selling the product, and more than half of the 15,000 sales teams have at least one Agentforce deal underway.
Benioff said during the earnings call last month that Salesforce has reached about 200 deals since it began selling Agentforce at the end of October, and is currently negotiating thousands of deals, although these deals have yet to have a significant impact on its revenue.
Adam Evans, head of artificial intelligence at Salesforce, highlighted the importance of AI agents by comparing them to new apps:
AI agents are “basic support technology for the robot era” and will penetrate various industries in the future. In the medical field, AI agents can act as immediate care assistants; in the retail sector, they can be personal shopping advisors; and in the financial sector, they can provide personalized services to customers as on-demand advisors.
“Requires a switch”
Looking to the future and achieving the ultimate goal, the reality may be more complicated.
A person doing business with Salesforce said that for any new enterprise software product, it usually takes 12 to 18 months for customers to go from trial to operation in critical business systems.
Cost is another issue. To use Agentforce, Salesforce recommends that customers rent Data Cloud, a service that stores the files Agentforce uses in the decision-making process. Data Cloud's pricing is based on the amount of data customers store, and it's easy to cost more than $0.1 million per year, even for smaller companies.
Salesforce charges $2 per conversation for the Agentforce customer service version, which is much more expensive than similar products from customer service rivals Zendesk and Intercom.