In a Dreamforce presentation, senior Salesforce leaders Jenny Tran, Amar, and Taci Parker introduce AgentForce, the company’s internal AI‑agent platform built on a trust layer and five core attributes (role, instructions, data, actions, guardrails/channels). They explain the "customer zero" approach of piloting these agents within Salesforce before external release. Three flagship agents are showcased: (1) Lead Nurturing Agent, which automatically contacts low‑scoring inbound leads with personalized emails and meeting scheduling, reaching over 80,000 leads in three months and freeing SDR capacity; (2) Sales Coach Agent, an on‑demand AI coach that lets sellers rehearse pitches and negotiations, providing data‑driven, personalized feedback to boost deal velocity and win rates; (3) Sales Agent (the "hero"), which aggregates structured and unstructured data from Sales Cloud and Data Cloud to deliver competitive, pricing, and customer insights, executive briefs, and onboarding support, cutting research time from hours to minutes. The speakers share operational details—prompt customization, data grounding, channel integration (Slack, mobile, CRM), and guardrails—and report key metrics: 76% of seller time previously spent on non‑selling activities, a 24/7 assistant for SDRs, and significant productivity gains. They conclude with five implementation lessons: start small, iterate and tune prompts, prioritize data quality, measure multi‑dimensional impact (adoption, efficacy, performance, value, job satisfaction), and build blended teams of business, tech, and users.
Chapters
Good afternoon. I'm Jenny Tran. I'm a senior product director at Salesforce. I've been at the company for eight and a half years now, but I've been in the ecosystem for more than a decade and this is my 10th Dreamforce. It's super special for me. I get to lead teams of Salesforce product managers and admins and developers who really help focus on driving seller productivity for our 25,000 salesperson organization. And this is my leader, Amar.
Hello everybody. This is Amar. I lead the sales and partner technology here at Salesforce. Been here 12 plus years. Been in the industry for more than 25 years and I can say with full confidence this is the best time to be a technologist and the transformation that we're going through. Love to share the journey today with you guys. Taisi.
I'm Taci Parker. I'm their business partner. My team is business architecture and governance and we enjoy leading our own internal solutions to our sellers. I've been trailblazing for 21 years, the last 10 of them here at Salesforce.
So before we dive in, I just want to explain really quickly what it means to be customer zero at Salesforce. It means that we're the first to use to test and pilot our own products before they reach all of you, our customers. It means that our sellers get to make the product better again before we bring them to market for all of you and that you can trust that the products are scalable and reliable and, and ready for use. And this is our agentic enterprise. So these are some of the top agentic use cases that we have here at Salesforce. We have agents that are improving the employee experience. We have agents that are driving operational efficiency. We have agents that are helping our customer engagement. On our websites like on help.salesforce.com and on salesforce.com we have agents that are helping our sellers be as productive as possible to drive growth and profit for the company and to focus on our customers. All of you. And this is super exciting for us, the momentum that we have internally because this is truly changing the way that we're working internally at Salesforce and it really did require a lot of strategy and a lot of focus up front. So before Tacy dives in to our sales agents, I want to explain really quick the five attributes that make up an agent. First is the agent role, and the role is the topic and the jobs to be done that the agent can perform within an organization. You'll then have instructions that the agent is going to follow so that it can Understand how to perform that role for you. You then feed it through. Feed it the data. So we fed the agent unstructured and structured data within our data cloud so that the responses are accurate and reliable. You then assign an action so that the agent can take actions for you. And then there's also guardrails where you can explicitly say what the agent should be doing or should not be doing. And the channel is essentially where the agent can show up. So these are the five attributes that make up an agent. And this is all built on the trust layer. All right, so with that, I'm going to hand it over to Tacy and she's going to talk a little bit about some of our sales challenges and how we identified our sales cycle.
Thanks, Jenny. So 76% is the amount of time that we identified our sellers were spending on non selling activity, which is a big problem. We have over 15,000 sellers here at salesforce, so there's a huge opportunity for us to turn around that productivity loss using agentforce. So this is our agentic sales cycle. And in this new sales cycle, we're transforming our lead to cash processes by figuring out which of the tasks an agent can handle and then giving sellers back the time that those agents are taking in those jobs. I'm going to show you three agents that we're leveraging in the first phase of that cycle. We're addressing more leads than ever using lead nurturing agent. Huge productivity boost to sellers using sales agent. And then we're preparing our sellers for customer conversations using agented coaching. So let's hop around and take a look first at the lead nurturing agent and what we've done with it to tap into our leads. 25% of our overall inbound sales is all that was getting directed to our sales development reps. So with that, we're leaving a lot of opportunity, a lot of revenue on the table. And so the lead nurturing agent is able to go in and action on all of those leads. We're going to do a quick demo of it really quick.
In the past, our US sales development reps were only able to follow up on one in every four leads. The other 75% scored too low to justify human outreach. And we knew that was a problem. It meant we were missing opportunities and letting revenue slip away to solve this. Agentforce for lead nurturing now automatically nurtures low scoring inbound leads around the clock with personal personalized email outreach and meeting scheduling so that no leads are left behind. The journey wasn't perfect at first. We started with our out of the box email prompts. But we quickly realized the agents weren't generating specific enough responses. A lead would ask about a certain product and the agent would just give a generic reply. To fix this, we customized our prompts to be more detailed, grounded in data from our own sales cloud and data cloud. This helped us ensure AgentForce acted as a true trusted extension of our ales team. Today, AgentForce is a 24. 7 assistant for our SDRs. When a low scoring lead comes in, AgentForce automatically sends a personalized email based on their product interest. It also follows our rules of engagement and opt out preferences, acting as a true member of the team. Once a lead shows enough interest, AgentForce sends a link to schedule a meeting and seamlessly passes them to a human sdr. The result? We've transformed our problem of too many leads into a great opportunity. Our reps can now focus on the leads that are truly ready to buy. And AgentForce ensures every other lead gets the attention it deserves.
So that's pretty cool. And actually, in just under three months, we've been able to approach over 80,000 leads through this agent. That would have cost us hundreds of SDR staffing to go after them with human effort. So it's amazing that we've had that impact. As Jenny mentioned, there are some key components that make up each agent. So let's check out what this agent has behind the scenes. First of all, its role is to be an outbound agent. So it's sending an email with personalized information to the prospect so that it can be nurtured, then leveraging data via data cloud that the customers provided to us via the website, as well as structured data from our sales cloud. And then it takes action. It is going to offer to schedule a meeting and then pass the conversation over to the seller. So now that we've engaged the seller with the prospect, let's look at how we're preparing our sellers for their first conversation with Sales Coach. So Sales Coach is a dedicated coach. We put in every single seller's pocket. It helps them prepare, it allows them to practice pitches as well as role play negotiation in a deal. Take a look at a demo.
Let's dive in into a demo to see how Agent 4 sales coach is helping our salesforce sellers. I'm going to play the role of an AE and I'm preparing for a meeting with my account Omega who is really interested in learning more about AgentForce Sales Coach. Now I'm ready to start refining my pitch for this opportunity and I'd love feedback on this critical deal, but my leader is away at a conference this week. Fortunately I have agent 4 sales coach that can help and I can launch it directly from the Opportunity Record Insight Salesforce to practice my pitch. Hi, since our last conversation I think a lot about how we help you with your search online. Would you want to see a demo or learn more? Now Agent 4 sales coach is processing my performance to provide individual, personalized and actionable feedback using details from the account and opportunity records in our CRM. Now here I see my feedback and includes a deal summary, feedback on my skills, my knowledge of the product and even additional recommendations and next steps. All of this will help me feel more prepared and confident so I can ace my upcoming meeting with Omega. It also tailors the feedback on things most important to our company's goals. In this instance, AgentForce. Now all of this is available out of the box and can also be customized to add in more specifics. With AgentForce our sellers can practice their pitch or role play to sharpen negotiation skills with personalized unbiased coaching on demand.
All right, Sales Coach is super easy to use. 1 and 2 it's built on out of the box actions and prompts so we were able to put it in place very quickly within our company. And then what we learned most through this is that like coaches have to be really trustworthy to be effective and so in order to make the coach accurate we had to spend extra time monitoring and tuning up those prompts and being sure that the responses were proper. Practicing through a coach has driven increased deal velocity and win rates for the sellers that we've seen actually adding the coach to their pre prep routines and so it's been excellent here. Finally I'm going to talk to you about our sales agent which is one that we call a hero. And the sales agent is augmenting our sales teams by taking on tasks that are non selling oriented or repetitive. It's also serving up great information to sellers so that they can come to their calls informed and ready to go. The challenge that we are setting out to solve is pretty obvious. It Was that selling time 76% on non selling activities is crazy. So we went right after how we could reduce the research time which seemed to be taking up the majority of that effort for them and pull together all of the disparate content that was required to whether it was product competitive, pricing information or even just customer insights. This agent is able to pull them all together using multiple data sources, both structured and unstructured and package them up to give the Seller guidance right there where they're at. So let's take a quick look at how the sales agent is shaving hours out of that work for our sellers.
Giving our sellers actionable data in a presentable format can be a time consuming process. At least it was until we started using agentforce to better equip our sellers. Our sellers have started using Sales Agent, which can be launched via Slack CRM or on mobile, to get answers to questions, get onboarding support and create executive briefs. And we're seeing incredible results. With Sales Agent, we've turned questions that can take hours to answer into ones that are resolved in minutes, giving valuable time back to our sellers. Instead of our sellers digging for information, now, they can ask Sales Agent to deliver that to them right when and where they need it. For more tenured reps, time is often wasted trying to locate the latest information on competitors and product pricing. With Sales Agent, competitive intelligence and product details are available right at their fingertips by simply submitting a question to the agent. To do this, we prioritize the agent's topics and skills, which are specific tasks the automated agent can handle based on their impact on seller productivity and alignment with our revenue goals. Sales Agent also helps speed up customer and internal meeting prep. With a simple generate an executive briefing doc for me prompt, we're able to reduce the amount of time to prepare an executive brief or a quarterly business review deck by an hour. We've learned a lot from our sellers, adapting the agent to work more intuitively and better assist them. Now that you've seen how AgentForce empowers our sales teams, imagine the impact it can have on your own organization.
With that, I'm actually going to fast forward here through some of our success metrics so you can just see them in a blank. We've talked a little bit about it and hand it off to Amar so you can share with us what we've learned and what you suggest we do next.
The five lessons right. Number one is start small, go fast, avoid applying AI everywhere all at once. Don't get caught up in the whole excitement. We did the same mistake and we corrected it and that's where we really targeted the key use cases, the real impact drivers. And that's what we call hero agents. You saw a few of them today, Sales agent as well as the sdr. They're great examples of where we are driving value. Number two, you have to think about launching an agent is the beginning. It's a lot of iteration, it's a lot of tuning. You'll not get it right in the first time, obviously, because you don't, you know, it's behavior changes. It's a dynamic environment. Agents adapt and they learn and they reason. So definitely this is a big, big learning for us and we continue to invest in it. Number three, data we know is critical, but right. Data is gold. You have to prioritize quality over quantity because you're training agents as your new hires, and you have to make sure that you're incrementally building up your assets. So keep investing in data. Data360 has been a game changer for us. Right. Number four, measure value and impact. Again, we tend to measure who's using the agent. That's not enough. You really need to go beyond that and really look at impact from all dimensions. And then finally, the fifth one, the most important. This is a new ball game. You have to work differently. And it all starts with the mindset of the people. There's no it. There's no business. You see all of us on the stage. It's one blended team. And we are also having our end users in the process all through the thing. And you have to reimagine the process. You're not automating a legacy process. You are reimagining it from how humans and agents can work together. That's the only way you will actually end up really driving that value. This is what I mean by measuring. You have to go a little bit beyond just the activity. You see these five dimensions that we use, it's about adoption, it's about efficacy, it's about performance, it's about the value, the actual critical measure that you're doing. And finally, it is about job satisfaction. You're basically having humans interact with agents. And then finally, this. This is another framework that I would love for you guys to take back. This is the blended team. It's users, business, and tech all together throughout the life cycle. Thank you so much.
