Salesforce introduces new AI agents to automate sales, support tasks
SiliconANGLE Maria Deutscher ● Covered by 3 sources
Salesforce rolled out new AI agents for sales, support and internal help desks. The big twist: one of them can plan across long sales cycles and reuse data between chats.
Based on reporting by SiliconANGLE, Maria Deutscher — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Salesforce has added a fresh batch of AI agents aimed at two of the most tedious corners of enterprise software: chasing deals and handling support. The company said most of the new features are available now, with the rest due by year’s end, and they arrive alongside an updated version of Agentforce Coworker, Salesforce’s automation tool inside several cloud services.
The sales side gets the most attention. Hunter is built for B2B sellers, helping them find leads, write outreach emails and handle the messy steps that follow. Salesforce says it can also prep presentations and draft proposals. What makes Hunter stand out is a feature called the long-horizon runtime, which lets it keep track of work over longer periods and carry data across chat sessions. For now, that runtime is only in Hunter.
Hunter isn’t alone. Piper can greet B2B prospects on a company’s website and route them to the right salesperson. Carter is aimed at online retailers, answering product questions and adding an in-chat checkout widget. On the support side, Casey can process product returns, while Fin handles similar customer service work. Fin uses technology Salesforce got earlier this year in a $3.6 billion acquisition.
Salesforce is also pushing deeper into internal operations. Paige is meant for employee HR and tech support requests, and Marshall is there for repetitive supply chain tasks. Companies can customize all of this with Agent Script, which Salesforce says lets developers force agents to follow specific instructions and reduces hallucinations. The update to Agentforce Coworker adds Multi-Agent Orchestration, output-quality optimization for subagents, and AI Skills to make customization easier. In other words, Salesforce wants AI agents not just to answer questions, but to do the boring work in pieces and keep doing it.
My take — AI-written commentary, not fact-checked reporting
This is Salesforce doing what enterprise software companies always do once the AI buzz settles: wrap automation in controls, call it coordination, and sell it as productivity. The interesting part isn’t the chatterbot gloss; it’s the move to long-running tasks and stricter instructions, because that’s where these tools stop being demos and start being office machinery. The hype crowd will scream autonomy; the grown-ups will ask who gets blamed when the agent goes sideways.
Read more about this at: SiliconANGLE