Agentforce is reinventing how visitors navigate Salesforce’s websites.

When visitors couldn’t find what they needed, their simple questions turned into contacts and dead ends.

Every month, millions of people visit Salesforce’s websites to explore products, pricing, events, and company news. Until recently, those visitors had to manually sort through an extensive menu of dozens of primary products, search results, and a product documentation library to get the answers they needed. If they couldn’t find what they were looking for, they often filled out contact forms to request more info — leading to an influx of unqualified inquiries that created extra work for teams.

In 2019, Salesforce took a first step toward solving this with an Einstein chatbot on the homepage. Although it answered common product questions and helped thousands of visitors each month find basic information, the bot relied on scripted replies and couldn’t go beyond predefined flows. And because it was tied to a standalone knowledge base, company and product updates required time-consuming manual changes — making it hard to keep answers current.

Salesforce spotted an opportunity to offer a faster, more intuitive experience that would give every user the right information, right away. Their goal wasn’t just to improve search — it was to completely reimagine how the website guides, engages, and converts visitors.

“Visitors come to our website with a specific question or intention in mind,” said Dana Greenberg, Senior Director of Product Management. “With new technology at our fingertips, we saw a path to getting the most relevant information and actions to them right away in a way that traditional search or chatbots couldn’t.”

With Agentforce, visitors can ask natural-language questions and get immediate, conversational support — right on the site.

With Agentforce, the agentic layer of the Salesforce Platform, Salesforce is fundamentally rethinking how people find answers on their website. Instead of searching through menus or scanning pages, visitors can now ask Agentforce conversational questions — and get helpful, accurate responses in return.

For example, if a visitor asks, “What’s the difference between Sales Cloud and Revenue Cloud?” Agentforce uses the Atlas Reasoning Engine to understand the intent behind the question. Then, using retrieval-augmented generation, it searches over 2,000 web pages in Salesforce’s proprietary content management system and more than 400 records from product and customer catalogs to find the most relevant information — and turns it into a clear, direct answer in seconds. 

It’s all powered by Data Cloud, which unifies structured and unstructured content from across the Salesforce ecosystem into a single, up-to-date knowledge base. A content pipeline automatically maps metadata and content attributes on newly published material and feeds it into Data Cloud — so answers stay current without requiring manual updates like with the prior chatbot. This means better experiences for website visitors and a more scalable approach for the teams managing it.

Agentforce responses can also adjust dynamically based on what visitors share about themselves. For instance, if someone says they’re from a small business, the agent tailors answers toward small business-friendly solutions. If they’re in retail, it surfaces customer stories from companies in the same industry. This kind of context-aware reasoning helps users get relevant answers faster — and gives sales teams better-qualified leads.

“By asking contextual follow-up questions, it’s more humanlike,” said Sharat Radhakrishnan, Principal Architect at Salesforce. “Agentforce is a huge leap from the chatbots of the past. It delivers that logical next response that you don’t have to rigidly define.”

To ensure quality and control, Salesforce is also rolling out a new flagging system that lets its teams suppress specific pieces of content from being used to ground Agentforce’s responses. This gives content owners more oversight into what information the agent draws from — and helps prevent outdated or irrelevant content from being surfaced.

Agentforce doesn’t replace search or navigation — it just gives visitors another way to find what they need. “Eventually, the goal is to align all of these experiences so that you come to salesforce.com, and you can just find what you need in the fastest possible way – and in your preferred way,” said Charlie Harris, Software Engineering PMTS at Salesforce.

Since launching in February, Agentforce has handled more than 185,000 conversations on the salesforce.com U.S. website — and has recently expanded to most global Salesforce websites, too, thanks to its multilingual capabilities. Today, Agentforce supports conversations in 11 languages — including Finnish, Portuguese, and Thai — enabling more users around the world to find what they need in their preferred language. Digital teams across the company are also deploying Agentforce on leading brands owned by Salesforce, including slack.com, mulesoft.com, and tableau.com.

Agentforce-driven leads have generated a $20 million lift to marketing-driven sales pipeline.

Agentforce doesn’t just help visitors find answers — it helps Salesforce’s sales team find stronger leads and reduce time spent on unqualified ones.

When a visitor asks about pricing — whether they type “Data Cloud cost” or “How much is Marketing Cloud?” — Agentforce pulls the latest information from Salesforce’s product pricing catalogs. Instead of a plain text reply, it displays dynamic cards with plan details and feature breakdowns, which often include buttons with relevant offers like a free trial. When a visitor clicks one of those links, they’re prompted to enter their contact information to sign up. That data is automatically captured and used to create a new opportunity in Sales Cloud — giving reps the context they need to follow up with a well-qualified lead.

“We’re excited about the possibilities for Agentforce to reason the best response or offer to put in front of visitors,” said Greenberg. “This is a whole new way to experience our website and content.”

If a visitor does want to speak directly with someone, Agentforce makes that easy, too — no need to fill out a form or wait for an email follow-up. Agentforce transfers the conversation, along with a summary, to a sales development rep (SDR) in real time, delivering a smooth handoff that helps qualified prospects get answers faster.

In the legacy chatbot experience, questions about product information and pricing were always routed to SDRs to answer. With Agentforce answering these questions, SDRs save 150 rep hours per month so they can prioritize stronger prospects. That focus is paying off: In 2025 alone, Agentforce has powered over 30,000 new leads. The likelihood for a contact who engaged with Agentforce to convert to a lead is 1.8 times higher than the website average, leading to an incremental $600,000 increase in marketing-driven pipeline per 1,000 site visitors from Agentforce. 

Not only that, but Agentforce-driven leads convert to opportunities that are 90% more valuable than the average for Salesforce web leads, generating a $20 million lift in marketing-driven pipeline.