Salesforce News & Insights

For a technology first popularized less than a year ago through an open-source project called OpenClaw, personal agents are rapidly hitting their stride. Instinct AI, a hugely popular invite-only personal agent, already boasts close to $1 billion in annual transactions, according to its founder, despite having no dedicated app — users simply text or call it.

For businesses, the opportunity is clear and immediate. Personal agents shopping on behalf of consumers, booking reservations, managing returns, and resolving customer service issues have the potential to drive substantial new revenue and a tremendous uptick in business-agent interactions. But the risks are equally apparent. If not properly governed, a massive influx of autonomous interactions could harm customers and businesses alike.

So, is there a safe way for businesses to allow personal agents in without losing control? In a new article, Mark Wakelin, EVP, Forward Deployed Engineering, Salesforce says that answer is yes, but only if those personal agents are met with business agents to manage the interaction and own the relationship.

salesforce.com/news/stories/p…

Today's AI agents are powerful tools for businesses looking to better connect with their customers. They can hold conversations, reason through multistep problems, take independent action across systems, and more.

But large language models are probabilistic in nature and excel only with the right context, data, and instruction. Otherwise, they can deliver incorrect answers and even take incorrect actions.

Rather than leave businesses to decide between AI agents and an old-school, purely deterministic approach, which keeps things safe by adhering to strict rules but offers a limited range of abilities, Salesforce combines them.

“Agentforce splits the agent brain into two layers,” said Kathy Baxter, Salesforce’s Principal Architect of Ethical AI Practice.

Baxter shares how to combine the strengths of these dual approaches, how humans stay in charge, and why testing before launch is only the start:

salesforce.com/news/stories/g…

A Salesforce customer since 2014, Xero - a global small business platform - has recently launched a conversational customer service agent.

“We went live in April to all 5 million of our customers,” said Nigel Piper, Executive General Manager, Xero. “[And] what we were able to do with our service agent is to bring channel choice into the chat experience. So if someone wants a call back, if they want to deal with a live chat interaction, if they do want to just raise a case, they don’t have to leave the agent. We can give the choice within the agent. They can just click a button. We’ve captured all the information that the customer has already given us, and we can use that for a better experience for our customers.”

Learn more about how Xero is redefining channel choice with AI agents.

A study by NVIDIA revealed that an agent can correctly identify a request as insecure in its internal “thought” process and yet proceed to execute that harmful action anyway.

So when AI agents dynamically connect to external tools, this “thought-and-action disconnect” can cause an agent to execute harmful function calls or pass insecure parameters to third-party tools. In multi-agent systems, that kind of failure can trigger cascading breaches across connected third-party platforms.

This phenomenon highlights a critical reality: one agent might be reasoning about being safe, but that doesn’t guarantee it is safe.

The key to fully mitigating this risk? In a new byline, Kathy Baxter, Principal Architect of Ethical AI at Salesforce suggests a hybrid approach — combining the LLM’s probabilistic reasoning with hard-coded, deterministic guardrails that exist outside the agent's internal reasoning loop.

salesforce.com/news/stories/t…

Today’s AI agents are powerful tools for businesses looking to better connect with their customers. They can hold conversations, reason through multistep problems, take independent action across systems, and more.

But large language models are probabilistic in nature and excel only with the right context, data, and instruction. Otherwise, they can deliver incorrect answers and even take incorrect actions.

Rather than leave businesses to decide between AI agents and an old-school, purely deterministic approach, which keeps things safe by adhering to strict rules but offers a limited range of abilities, Salesforce combines them:

1️⃣ The probabilistic model handles language and reasoning and predicts the most likely answer to a question.
2️⃣ The deterministic layer includes preset deterministic business logic, workflows, and permissions, so every action or response is subjected to a company’s prescribed policies, procedures, or escalations to humans.

Kathy Baxter, Principal Architect of Ethical AI at Salesforce explains how pairing a probabilistic model with deterministic logic gives agents the ability to take action — without giving up the guardrails.

salesforce.com/news/stories/g…

Southwest's AI journey started out with a single agent to answer flyers' questions. The company is now moving to a multi-agent architecture to increase its customer service experiences.

Rather than customers having to conduct several different searches for information, they can ask all of them in one place and the agent will be able to assist, from name changes to flight booking adjustments, refunds, and more.

"[AI agents] allow more time for our human representatives to assist with more of the customer situations where we're needing more hospitality and empathy," said Megan Rauber, Manager, Customer 360 at Southwest Airlines.

Learn more about how Southwest Airlines is using AI agents to change the way people fly:

Today, most AI and agent use isn’t shared. One person, one screen, one agent in a private tab, solving one problem. The person and the agent both learn something — and then the project ends, they move on, and the rest of the company never sees any of it. The knowledge stays in the tab.

AI is worth far more to a business when it works out in the open, in the same places people already work. That’s the idea behind this next era of Slack: share context, build on it together, and generate more context. That loop is how companies move faster and make AI work at real scale.

Rob Seaman, Senior Vice President of Enterprise Product, Slack shares more about the shift from smarter individuals to a smarter company.

salesforce.com/news/stories/m…

When customers need support, they expect to have straightforward conversations, in natural language, that result in a quick solution.

For Canon, customer inquiries may be as simple as asking what ink is needed for a certain printer. Yet, with the variety of models and SKUs, the answer is often complex. Agentforce was the key to meeting customer expectations, reducing friction, and ultimately responding to these inquiries faster. The company has been able to deflect 69% of customer issues with Agentforce.

"I had my team go through and manually verify the data that we were seeing was correct, because I didn't trust it," said Bill Duval, VP, Technology & Information Management at Canon.

See how the power of Agentforce has been a game changer for Canon:

The Formula 1 fanbase has grown over the past five years to 831+ million fans around the world. With that growth, fan needs and interests have shifted.

"Part of our journey with Agentforce was very much around, how can we use this to help our fans, our new fans who perhaps don't have as much knowledge of Formula 1 or motorsport," said Claire Sparks, Head of Portfolio and Technical Initiatives at Formula 1. "Marshal will be able to help our fans on our web and app channels through conversational questioning and responses to really just help them find out more about what's important to them."

Learn how Formula 1 uses AI to educate a new generation of fans:

Last week at Dreamforce 2026, the conversation buzzing among executives, customers, and trailblazers centered around the next era of AI. An era where the creative, probabilistic reasoning of AI works with the deterministic rules of business.

"When you press the power button on your TV, it's always going to come on, right? Versus if you tell your child, 'hey, will you turn off the TV?' Maybe they do, maybe they don't," said Michael Andrew, Chief Data Officer at Salesforce. "Agents are the same way. You want to teach them, you want to monitor them. And then, as they prove more and more capable, you get them more controls."

When these systems unite across data, applications, agents, and interfaces, AI can unlock trapped value across the entire enterprise, at a scale humans alone could never match.

See what industry leaders on the ground had to say about the future of AI in the enterprise:

Dreamforce 2026 delivered a wave of announcements defining the next era of enterprise AI — from Salesforce’s own reasoning model, Koa, to new agentic capabilities spanning sales, service, government, and industry. And Salesforce customers are already seeing real ROI:

🔹The Grout Guy: By automating the admin behind scheduling, quoting, lead management, and technician preparation, the company has seen a 20% increase in quote conversion rate.
🔹AT&T: With "Agent Briefing" for sales reps, the time it takes to upgrade a phone in a store reduced from 30 minutes to 10 minutes.
🔹Box: Integrating Salesforce data and its own unstructured content directly into Slack has allowed deal teams to access comprehensive customer information without switching between different user interfaces.
🔹Engine: New reasoning models from Salesforce help the company handle complex, multi-system issues like travel disruptions and automatic dispute resolution.

Catch the essential rundown for leaders tracking where agentic AI is headed and what it means for the enterprise: x.com/SalesforceNews…

For most of Agentforce's history, one general model did the reasoning for almost every task: Intent detection, prompt-injection screening, toxicity, evaluation, reranking, etc.

Over the past eighteen months Salesforce has changed that, building a series of smaller models, each built for a specific job. But the company left all the core multi-step reasoning with the general model.

That changes with Koa, Salesforce's first AI reasoning model optimized for enterprise work, running inside the Salesforce trust boundary. Salesforce co-engineered Koa with NVIDIA, post-training their NVIDIA Nemotron 3 Super model for the most common enterprise tasks. Customers can now use Koa to run their AI agents' processes.

Koa is already running at Salesforce inside internal employee workflows, and is now moving into customer pilots. Early benchmarking on CRM specific tasks such as updating an opportunity, routing a case, or scheduling a follow-up are encouraging, outperforming the strongest general-purpose models.

In a new article on the Salesforce Newsroom, Jayesh Govindarajan, EVP, Software Engineering and Silvio Savarese, EVP and Chief Scientist, Salesforce AI Research share more about why Salesforce post-trained its own reasoning model.

sforce.co/4h0i9vu

CIO Reports: Salesforce launched AIforce, a live interface layer for enterprise software it hopes will transform the way knowledge workers access enterprise data.

By unlocking Salesforce data, workflows, business logic, semantics, permissions, security, and governance for any AI interface, the company also aims to broaden the scope of its business by reaching knowledge workers within the enterprise who’ve never used Salesforce before. The promise is the ability for users to create custom and composable applications that leverage Salesforce as the system of record.

Analysts felt the venture could signal a new agentic era for enterprise software.

“Core ERP, core CRM, core HR, most of these tools are going to see a similar sort of realignment where the models will provide probabilistic capability that uses the best of breed from your intelligence coupled with the rich, proprietary data that businesses depend on to operate,” said Daniel Newman, CEO and principal analyst of research firm The Futurum Group.

AIforce is launching with Claudeforce, Slackforce, and Agentforce Coworker (an AI teammate within the Lightning interface), but Stokes made it clear that more AI platforms would be announced soon.

cio.com/article/422218…

Dreamforce is in full swing, and the past two days have been full of innovations and announcements to unlock more value than ever across the Salesforce platform. Highlights include:

🔹 the introduction of AIforce, a live interface layer that brings the enterprise knowledge inside Salesforce to any agentic platform
🔹 the announcement of Koa, Salesforce’s first CRM reasoning model for Agentforce, built on NVIDIA Nemotron, and developed through deep technical collaboration with NVIDIA
🔹 an expansion of Live Nation’s use of Agentforce, bringing 24/7 fan support to Live Nation venues across the U.S.
🔹 a new Missionforce partnership with OpenAI to bring its leading AI models to secure government environments

and more.

Follow Salesforce News & Insights to stay up to date, and check back for more updates as they land ⬇️

bit.ly/3SOBDdg

An Australian grout company has harnessed Salesforce and AI to help scale business nationally.

The Grout Guy is using agents to automate the admin behind scheduling, quoting, lead management, and technician preparation, allowing their team to spend more time looking after customers. Since implementing Salesforce, the company has seen, on average, a 20% increase in quote conversation rate.

Anthony M., CTO of The Grout Guy, shares how they became an AI powerhouse:

One year ago Salesforce launched Missionforce, its agentic platform for government that brings trusted AI, data, apps, and cloud technology to the government’s most critical missions. Since then, Public Sector has become Salesforce’s fastest-growing industry business at scale, with its Defense sector business growing 80% year over year as demand accelerates.

Today, Salesforce supports some of the government’s most complex and consequential missions across more than 30 countries, 60 U.S. federal agencies, all 15 U.S. Cabinet departments, all 50 states, and all six branches of the U.S. military.

From citizen services and agency modernization to military readiness and national security, government organizations have massively expanded their use of Salesforce over the past year:

➡️ The U.S. Army awarded Salesforce a 10-year Indefinite Delivery Indefinite Quantity (IDIQ) contract with a $5.6 billion ceiling to expand Missionforce National Security across the Army and Department of War, helping accelerate procurement, connect fragmented systems and data, and improve operational readiness.
➡️ The U.S. Army Human Resources Command (HRC) selected Missionforce National Security to deploy AI-powered HR support for 9.2 million soldiers, veterans, civilian staff, and military families.
➡️ The U.S. Air Force 441st Vehicle Support Chain Operations Squadron (VSCOS) deployed Missionforce National Security to modernize sustainment and operations for its $13.5 billion fleet of over 84,000 vehicles across nearly 389 global locations.
➡️ U.S. Department of Transportation (USDOT) is expanding its agencywide modernization with Missionforce, including unifying more than 10 systems on one platform and deploying Agentforce to support citizen services, grantmaking, and transportation safety.
➡️ Transportation Security Administration (TSA) deployed Ace, an AI agent built on Public Sector Solutions, to help travelers navigate airport security.

Missionforce is also expanding its partnerships across the AI ecosystem to give government agencies greater choice and control over the models and infrastructure they use — including a new collaboration with OpenAI to bring advanced AI models and accelerated computing to support Missionforce workloads and secure government environments.

sforce.co/4yD789b

Salesforce today announced new Missionforce capabilities that give government agencies greater control over how they develop, customize, and deploy AI while streamlining policy execution, enhancing operations, and improving citizen support.

Salesforce also announced a new Missionforce partnership with OpenAI to bring its leading AI models to secure government environments. Salesforce's partnerships with leading AI innovators will bring together Missionforce capabilities with U.S.-origin models and trusted AI and mission infrastructure.

These partnerships will unlock immediate operational impact:

🔹 decision clarity for operators in the field
🔹 greater efficiency and mission-readiness for agencies
🔹 faster, more responsive services for citizens

“Government agencies want tailored AI that runs everywhere they operate while staying within the secure environments their missions demand.” said Kendall Collins, CEO of Missionforce & Government Cloud. “Missionforce enables this at scale, with purpose-built AI that executes the mission and enables agencies to solve their hardest problems while maintaining operational control.”

salesforce.com/news/stories/m…

We are in the midst of an interface revolution, and it is about to transform the enterprise.

Today at the Dreamforce 2026 keynote, attendees and online viewers witnessed an AI interface that is dynamic, intelligent, composable, and alive: "When you put all those things together, you get AIforce," said Marc Benioff, Chair and CEO, Salesforce. "You get those capabilities and you get this live interface that untraps the value."

Salesforce executives, partners, and customers joined Benioff to discuss the power of combining AI's probabilistic intelligence with enterprise intelligence. As Jensen Huang, President and CEO, NVIDIA said, "The sky's the limit for every industry, for every single country, so engage AI, don't get left behind."

Watch the keynote in under 7 minutes, and stay up to date on the latest news and stories coming out of Dreamforce here: bit.ly/3SOBDdg

Twenty years ago, Salesforce sparked an entirely new ecosystem made up of the people who build, customize, and govern Salesforce technology for other companies. At the time, SaaS technology was still in its infancy, and these administrators and developers came together to learn new skills, share best practices, and support each other in their career journey.

Now, with AI and agents changing work, companies are figuring out how to adapt yet again, and Trailblazers are leading that shift.

For two decades, this global crew of millions has powered the Salesforce ecosystem. They’ve built apps, customized agents, and brought Salesforce technology to life for businesses everywhere. And in doing so, they’ve shaped Salesforce products, supercharged careers, and driven real economic growth around the world.

"Trailblazers have guided businesses through every wave of innovation for 20 years, and now they're leading the way into the AI era," said Meredith Brown, SVP, Trailhead & Community. "We're grateful for this community, for how they show up, go first, and support an entire ecosystem of Trailblazers skilling up for what's next. They're not just part of the Salesforce story, they're the heart of it."

sforce.co/4AxvRxJ