Salesforce’s AI Research team used its latest roundtable to position ‘ambient intelligence’ as the next practical step for enterprise AI, and to show how that idea is moving from research into frontline workflows across sales and service
That matters in CX because work happens mid-call and mid-case. It also happens under pressure. Real-time guidance can lift productivity while reducing training burden. It can also tighten compliance if it captures what was said and turns it into structured next steps.
In the roundtable, Salesforce executives tied the concept to the idea that enterprise AI must operate as a system, not a single model. They argued that companies will need orchestration, memory, evaluation, and control layers to make agents reliable.
In an assessment, Silvio Savarese, Executive Vice President and Chief Scientist of Salesforce AI Research at Salesforce said:
“Enterprise AI is not just about having a strong model, it’s about building a system you can trust in real workflows.”
Inside The Demo: Ambient AI During Live Selling Moments
Salesforce used a live sales scenario to show how ambient intelligence could work inside meetings.
The demo centered on an in-call assistant that listens to the conversation, pulls out customer intent, and helps the rep move faster. Salesforce presented it as a way to reduce manual note-taking, make follow-ups more consistent, and keep the conversation anchored to the next best action.
For CX and revenue teams, this type of always-on assistance can also change how coaching works. If the system can spot moments that predict churn risk, buying intent, or escalation, leaders can build more consistent playbooks. It also creates a new governance question. Enterprises will want clarity on what gets stored, what gets summarized, and what actions are automated.
Agentforce Upgrades Focus on Control, Scale, and Proof Points
Salesforce also used the session to underline that it sees Agentforce as an enterprise platform, not a lightweight bot layer.
The company highlighted progress on the controls and visibility enterprises need when they deploy AI agents across teams. It positioned these improvements as essential for moving from early pilots to scaled adoption, especially in regulated environments and high-volume operations.
Salesforce also pointed to early enterprise outcomes as proof that these deployments can drive measurable results at scale. Yacov Salomon, EVP Engineering, Service Cloud at Salesforce said:
“Uber used Salesforce Media Cloud to scale its ad business to $1 billion in revenue in just two and a half years, and to deliver a 70% faster sales process.”
He also said Southwest Airlines deployed an Agentforce FAQ agent on its self-help portals and achieved 20% call deflection and up to 50% containment, while becoming Salesforce’s highest consuming Agentforce customer globally within 90 days after launch. Salomon added that Adecco Group is using Agentforce to transform hiring with AI agents, including personalizing job seeker engagements and reducing time to hire.
Why CX Leaders Should Pay Attention Now
Ambient intelligence is easy to dismiss as branding. The substance is harder to ignore.
If Salesforce can deliver workflow-native AI that consistently captures context and reduces admin load, CX teams could see shorter handle times, cleaner CRM data, and faster case resolution. But the bigger win may be standardization. The same guidance can show up for every rep and every agent, not just the top performers.
Salesforce framed those examples as signals that ambient intelligence can move beyond demos and into production value. In the session, Salomon highlighted Uber’s $1 billion Media Cloud revenue milestone and 70% faster sales process, plus Southwest’s reported 20% deflection and up to 50% containment from an Agentforce FAQ agent.
At the same time, enterprises will need to treat this as a governance program. Real-time AI in conversations raises questions about consent, retention, compliance, and the accuracy of automated actions. The vendors that win will be the ones that make controls and observability part of the product, not an afterthought.
The future of CX won’t be defined by who has the most AI features. It will be defined by which platforms can make AI feel natural in the moment, and safe at enterprise scale. Salesforce is betting that ambient intelligence, paired with more mature agent controls, is how it gets there.
Join the conversation: Join our LinkedIn community (40,000+ members): https://www.linkedin.com/groups/1951190/
Get the weekly rundown: Subscribe to our newsletter: http://cxtoday.com/sign-up
