Agentic AI Interoperability: The Future of Enterprise AI | Box & Seismic

Hi everyone, I'm Brian Lansman. Today we're diving into one of the most important frontiers in AI agentic interoperability. Since launching AgentForce last year and the leading AI agent ecosystem, Agent Exchange in March, we've seen thousands of customers enhance their use of Salesforce with powerful agents and partner built components. Our partners are Now Powering a $6 trillion agent economy by building and enabling a whole new category of agent first businesses. But now the next big question is how do we get these agents to work seamlessly and securely with external agents and enterprise data sources and tools? Today I'm joined by Gary Lerhop from Salesforce, Yashoda Bhavnani from Box, and Tom Strader from Seismic. Hey everybody, how's it going? Gary, I want to start with you. Why do companies need to be agent first? And what does it take to become an agentic enterprise?

Thanks, Brian. First of all, excited to be here with this group being agent first. The idea of agentic transformation is a huge imperative for organizations. Something like over 80% of CIOs saying AI will be as significant a business as the rise of the Internet was in the early 2000s. And that's because by building a digital workforce of AI agents alongside humans, organizations stand to realize unprecedented productivity gains and basically the creation of entirely new business opportunities. If you think back to that rise of the web metaphor I was just talking about, there's a really big parallel here. There's a huge difference back then from companies who just added a website to their business versus those that embraced how the web forced them to rethink the servicing of their customers, how their employees could work together in better new ways. So we're basically at the same type of inflection point moment today, but now with agent first. And so companies need to think about transforming their enterprise into an agentic enterprise. So what does that take? Basically, it's about embracing open standards to interconnect your systems and plug them into agent driven experiences. Standards like MCP and A2A are gaining traction, they're coming on strong and they form the backbone of interoperability between agents. Now with Agent Force 3 and this launch, we're delivering enterprise grade AI supporting those standards. And that enterprise grade comes from us building interoperability with things like governance, identity and control at their foundation. You add that to a thriving open ecosystem of partner solutions, agents and support tailored to customer specific industries. And what you have is Salesforce delivering a trusted, open and complete AI agent platform. That's what you need, that's what it takes Tom.

Seismic is a global sales enablement platform and you guys were very early adopter of agentic AI, delivering some of the first agentic solutions on our Agent Exchange. Can you share how Seismic is building on AgentForce and the advantage of building agentic solutions for the Agent Exchange?

Yeah, absolutely. Thanks, Brian. And I can't believe it's been almost maybe six months since we started this journey released a few months ago. And the traction's been really phenomenal. And really just building for AgentForce is different than anything. Kind of like Gary was saying, it's different than how you thought about business a year or two ago. And as we think about developing on the Agent Exchange and releasing those features that our customers need, we're really focusing on those pain points to solve for individual seller workflows and sales managers in a way that agents can really take over a very big chunk of a workflow. And so today we have about eight actions on the Agent Exchange, but we're adding more every week because something like 80% of seismic customers are already Salesforce customers. And as we transition from the era of CRM to the era of agentic CRM, it was obvious to us that Agent Force was the place that sellers were going to be. And so it's been a really exciting journey to see the pull from our customers who are all trying to change the way that they sell. Right. Similar to the Internet era. Right. We're now in the AI era where people are really looking at how do we transform our sales motion. And we're excited that Seismic is playing a key role there. And the value we used to provide for human agents, we're now providing to AI agents. And so it's an awesome time to be in the sales space.

That's great. Let's go over to you, Yoshioda, Another great company, great enterprise software company, Box. Can you talk a little bit about some examples of actions or agents that are accelerating how companies become agent first from your company?

Absolutely. At Box, we have over 100,000 enterprises that rely on us for content services and they rely on us to manage their knowledge work and also do it securely. We serve a majority of Fortune 5000 and we also serve startups across the globe. And we recently ran our State of AI report that revealed that 87% of companies are already leveraging AI agents and making them permissions aware is what really, really matters to them. And that's where Box is really leading the charge to empower two things. One is organizations to leverage AI and agents to power their knowledge work for individuals and also enterprises to power their workflows, making it seamless and obviously really, really secure through what we deliver at Box, which is permission, aware, AI agents and secure. So why does this all work? This works because over 90% of content is unstructured. And AI agents just thrive on this unstructured content. What we're doing in Box is three things. We're building an agentic platform so you can choose, build and securely deploy AI agents. Second, we're building AI agents right into the platform so you can extract metadata, search across your content, research Q and A, and even create content using AI agents. And finally, we're making sure all of this is integrated seamlessly. So Box has already built over 41 actions for Asian Force. This will soon include Box's AI actions for Agent Force. And we're also exploring additional product strategies around Built in Agent. And one of the main use cases we've seen, and there's so, so many, is something as specific as leases. We have companies that have large, large leases across the United States. And answering a simple question, which you may find as simple, such as when what are the leases that are expiring in the next 30 days is actually really, really hard. And in order to do this, this company deployed Box AI agent for extraction and was able to classify and get the right insights from hundreds of leases that are really complex with a lot of data so that they could do audits and answer these types of questions. And there's so, so many more use cases such as this. Specifically, when you think of the content that's in Box and the content that's in Salesforce, and having agents being able to go between these to answer questions and help our customers.

So you guys are both building agents. You know, Yashoda, you're building an agent. Tom, you're building an agent. You're both listing those actions and some of those agenta capabilities on the agent exchange. But you're both dealing an enormous amount of content. But your agents, they're not talking to each other. So how do we fix that? You showed us what are some of the most significant challenges that enterprises face when they start to adopt multiple AI agents. And how's Box thinking about managing this complexity?

Yes, our agents will talk to each other. And one of the most important things that customers need from all of us together as we work together, is making it possible for customers to seamlessly interact across different agents and making sure those interactions are secure. And data that isn't meant to go across these systems doesn't actually infiltrate into different systems, and that's exactly what we're working on. In fact, we're already integrated into A2A. We were a launch partner with Google and we're already a part of that space. We're also supporting MCP calls and in fact, we're bringing our MCP server to major registries and are excited to explore this opportunity, obviously with Salesforce to deepen our AI and agentic partnership forward.

Gary, going back to you now, can you share how we're thinking about that a little bit? How Agent Force facilitates interoperability and orchestration of agents across different tech stacks and how our open partner ecosystem plays a role in scaling Agent Force and really scaling digital labor.

Yeah, totally. I touched on this earlier, but with Age of Force 3, we're bringing secure support for open interoperability for standards like MCP and A2A. We're able to do that with Salesforce's new AI Agent Gateway. It supports both MCP and A2A in a way where customers can securely connect AI agents to enterprise systems and third party tools with a registry and policies that help govern their use. So along with that, we've got the newly expanded Agent Exchange, which makes it easy to discover and deploy trusted external tools, pre validated MCP servers and eventually external agents through A2A, allowing customers ultimately to fast track their agentic deployment and development. Agent Exchange is going to be the place where we're listing those trusted MCP servers from across the ecosystem and you're going to see that today through partners like Box and with others.

Tom, on your side, how are you thinking about interoperability? How will it help seismic deliver on this next wave of innovation? How's it going to support your business?

Yeah, absolutely. I think when we looked at AgentForce for the first time, I think we all realized that the ecosystem was going to be the thing that made agents successful because you'd had companies for a while that were building the agent for their platform in very bespoke ways. Right. But they were doing that and you weren't seeing these companies being wildly successful because they were only solving a very small problem. And so when Agent Force came out, we were excited to see that probably the biggest ecosystem platform on the planet really leaning into Agentic, because we saw the future of what selling could look like. And so we knew that it wasn't going to be just seismic that we'd have companies like Box and other companies in the sales space that were going to latch onto AgentForce. And by doing that, now a seller that has a very Complex job that spans across Outlook and Box and Seismic and Zoom info and you name it, they're interfacing with dozens of tools a day, hundreds of prospects and meetings to go to that they have a complex job and that you need a multi agent ecosystem to facilitate them doing that job in the best way. And especially as we start to look at some of those higher volume tasks that we think agents can completely transform the way that they're done, you have to have all those tools in your toolbox and the agent has to be smart enough to know how to orchestrate among them and know the nuanced differences between something like my lease agreements, which maybe are in Box, versus how to negotiate a lease, which maybe is in seismic, or the training for negotiating a lease being a different type of knowledge than your archival system. And so those nuances are things that agents need to understand deeply. And I think between the combination of A2A and MCP, we start to see that future becoming reality and frankly, as seismic. We're very glad to see Salesforce and other large companies pushing the envelope on standardization on MCP and A2A, because a company like Seismic, we looked at them, right, but we weren't going to make the decision to say, hey, we're going to provide an MCP server or we're going to provide a 2A. But when companies like Salesforce and Box can make those investments and show the market that that's where things are going, it opens the door to all of these SaaS, vendors and even like private independent third parties and companies themselves to really invest in those standards. And it's a win win for everybody. So we're very excited about where the future is going. We're excited for agentic workflows, especially the fully autonomous ones. And we see that A2A and MCP are really the way that Seismic and other modern vendors are going to be able to provide the same value proposition they've always provided, but to the next generation of the workforce, which is agents. And maybe you heard it here first, but I call it Gen Agent. That's the next generation of the workforce and I'm just excited to be part of it.

Well, we're excited to have both of you as amazing partners and amazing part of our Agent Exchange. We're also beyond thrilled to have so many amazing partners joining us and becoming a great destination for all of this amazing interoperability. Same way we were when we launched the App Exchange back in 2006. Twenty years later, we're leading forward with the Agent Exchange here. At Salesforce. So that's all the time we have for today. Before we close it out, I'd like to invite our audience to check out the latest innovation on Salesforce's Agent Exchange and for all of our partners to join our webinar in a few hours at 12pm Pacific Time. Dig deeper into what Agent Force Updates mean for you. Registration link is in the comments below and I want to thank Gary Yashoda and Tom for joining me. Thanks to all of you for tuning in and have a great day.