What's Up with Tech?
Tech Transformation with Evan Kirstel: A podcast exploring the latest trends and innovations in the tech industry, and how businesses can leverage them for growth, diving into the world of B2B, discussing strategies, trends, and sharing insights from industry leaders!
With over three decades in telecom and IT, I've mastered the art of transforming social media into a dynamic platform for audience engagement, community building, and establishing thought leadership. My approach isn't about personal brand promotion but about delivering educational and informative content to cultivate a sustainable, long-term business presence. I am the leading content creator in areas like Enterprise AI, UCaaS, CPaaS, CCaaS, Cloud, Telecom, 5G and more!
What's Up with Tech?
How AI is Making Every Seller Your Best Seller
Interested in being a guest? Email us at admin@evankirstel.com
What would your sales team look like if every rep could sell like your top performer? That's the transformative promise of AI in sales, and it's happening right now.
Arjun Pillai, serial entrepreneur and CEO of Docket, an AI platform transforming B2B sales, reveals how his third venture in sales technology is fundamentally changing how companies approach go-to-market strategies. The problem? A staggering 90% of sales knowledge exists either in people's heads or buried in unstructured data like call recordings and chat conversations. This creates endless inefficiencies, slows down new hires, and leaves revenue on the table.
Docket's solution is brilliant in its simplicity: create a comprehensive "sales knowledge lake" that captures both structured and unstructured data, filters out the noise, and learns specifically from top performers. On this foundation, they've built two game-changing AI agents. Their AI seller autonomously engages website visitors with human-like conversations, while their AI sales engineer – the true breakthrough – provides real-time support during sales calls, helping representatives instantly answer technical questions or handle objections exactly as the company's best salespeople would.
The results speak volumes: customers report 12% higher win rates, 10% shorter sales cycles, and 15% faster onboarding for new reps. With over 15 months of enterprise deployment and 55,000+ questions answered, Docket has proven these systems can work at scale without the hallucination problems that plague many AI implementations. Through confidence scoring, reference links, and customizable guardrails, they've built the trust necessary for widespread adoption.
Perhaps most fascinating is Arjun's perspective on the future. While fully autonomous AI sellers will increasingly handle simpler, high-volume transactions, human representatives will remain essential for complex enterprise sales. The companies that thrive won't be those that replace humans, but rather those that strategically augment their capabilities – enabling every seller to perform like their very best.
Want to see how AI could transform your sales organization? Discover what's possible when you unlock the hidden knowledge in your go-to-market teams.
More at https://linktr.ee/EvanKirstel
Hey everybody, super excited for this chat. Today we talk about AI meeting sales and sales tech with an innovator in this space at Docket Darjan. How are you?
Speaker 2:I'm good, Evan. Thanks for having me Excited to have this chat.
Speaker 1:As am I. As someone who's been in sales for gosh 35 years, it's great to see this revolution happening. Before that, maybe introduce yourself what is Docket and what's the big idea behind the company.
Speaker 2:Yeah, absolutely. Docket is my third company. I have, fortunately or unfortunately, been in the sales tech and market space for the last 13 years. First company was a sales intelligence company that I sold to Full Contact. Second company was a sales intelligence company that I sold to Full Contact. Second company was a conversational marketing and sales a little bit of an ABM platform that I sold to ZoomInfo Now it's called ZoomInfo Chat and I was at ZoomInfo as the chief data officer for a little over two years and then I resigned and started Docket end of 2023.
Speaker 2:At Docket, I'm the co-founder and CEO as my title, but what we believe as a company is that if you take any company, less than 10% of the go-to-market knowledge is actually documented.
Speaker 2:The other 90% is either in somebody's head or lying in unstructured data like a call recording or Slack chats and Microsoft Teams chat and things like that. But this data is so critical for both the sellers and buyers. So what we do is, when we work with our customers, we help to bring together the structured and unstructured data, clear out the noise, learn from the best sellers and build it into what we call a sales knowledge lake for that particular company and then, on top of that sales knowledge lake. We have two agents that we give to our customers. The first one is the AI seller agent. That is an autonomous agent that can interact with your website visitors in a personalized way, human-like way, on the website, on the website. And then there's the AI sales engineer. That is an enabling agent that makes your sellers 33% more productive by giving them the help like a sales engineer would do.
Speaker 1:Wow, the AI sales engineer sounds fascinating. I started my career as a sales engineer, so maybe walk us through what it actually does and how it works day to day with the sales teams.
Speaker 2:Yeah, absolutely so, sales engineer. Fundamentally, a sales engineer does a few things right. They have the ability to ask deep discovery questions. They have the ability to answer complicated technical questions. They fill out the RFPs security questionnaires and they do the demos. Those are at a very high level.
Speaker 2:Obviously, it changes from company to company a little bit depending on the complexity of the solution that they are selling, but this is at a very high level. Obviously, it changes from company to company a little bit, depending on the complexity of the solution that they are selling, but this is at a high level, what they do. So what we do is, once we pull together this knowledge in the company into the sales knowledge lake, this sales engineer will have the ability to answer questions from the account executives, and that can be on a Slack. That can be when they are receiving an email. So they are in chrome. They receive an email our plugin pops up on the right hand side and give the answer to the to the email. The interesting thing is every answer we give is the answer that one of the best sellers in the company has given in the past.
Speaker 1:Oh, wow so yeah.
Speaker 2:So even a new account executive, who does not know all the details of the products or how to answer an objection or a competitive intel question, they'll be able to respond exactly like their most inward account executive. Now, one thing that we have recently launched is our ability to actually join calls in real time. So, evan, imagine you and I are on a call. You are the buyer, I'm the seller. You ask me a question. I have no idea how to answer. My options are either I say let me come back to you or I give you some BS. Both happen all the time on sales calls. But with Docket what happens is you press the shift button and the answer pops up in one and a half seconds on your right hand side. Wow, so the buyer wouldn't. You wouldn't even know that I have something on the right, but I get like short formatted text that I can quickly read and say yes, we do integrate with that system. And here is how it happens.
Speaker 2:That sounds amazing.
Speaker 1:You also talk a lot about your sales knowledge lake. What is it? How does it work? How is it different from your typical CRM or sales tool that's out there today?
Speaker 2:Yeah, core understanding there again is that for an account executive to execute their tasks they need the deal intelligence and processes that you think about a mid-take, mid-pick, or the sales methodologies entering details into a CRM, following up at the right time. All of that I would call as the deal intelligence. And then there is the product intelligence, which is about the product knowledge objection, handling, competitive, the enablement stuff that you typically get. So if you look at a typical CRM solution or any of the conventional sales tech solutions, deal intelligence is what you handle. If you go to salesforce or a gong, what you have there is you know what is the deal value, who is the account owner? Is it an enterprise customer? You know, um, what is the risk signals in the deal. All of that super critical, but that's what they cover. What they do not have is the product knowledge, because product knowledge is sitting in sales enablement tools and marketing collaterals and website and help center and RFPs that they have filled product documentations.
Speaker 2:That's what we pull and bring it into the sales knowledge lake. So, from a source standpoint, bringing together this siloed product and go-to-market knowledge into one place is very critical for any company to get started with AI, in our opinion. Then the second thing that we do is imagine a call recording Evan, we talk about what is the weather in Austin, what did you do this weekend? Good questions on a sales call, but not knowledge essentially. So you have to tune out all of that noise, because garbage in, garbage out. We want to keep the garbage from going in, so we tune out the noise. So we have a proprietary data cleansing process.
Speaker 2:The second thing that we do is you tell us who are your best sellers and subject matter experts. We go look at the interactions that these people have done in the last year their calls, documents, questions, answers. We weigh their answers slightly higher than the rest of the tribal knowledge, because of which our system becomes super accurate and learns the verbatim of the best sellers. The third thing is we understand how many products do you have for every product? Who are the competitors? What are the discovery questions? What are the objections. That's kind of like an onboarding thing that we do and with that we build the knowledge graph that is curated for your company. So I hope you get the picture. It is the deal intelligence and product knowledge and the three levels of pruning and reasoning that we do on top of that data. That makes it the absolutely pristine go-to-market knowledge graph that any company needs to execute. On AI.
Speaker 1:Well, so it sounds so impactful, but what so? What are customers telling you? Uh, the anecdotes or stories, feedback on how you're, you know, changing the sales game, basically.
Speaker 2:Yeah, absolutely so. It's a we see from a metric standpoint, we impact the win rate and sales cycle and onboarding time. We typically increase the win rate by about 12%, we reduce the sales cycle by about 10% and we onboard the account executives about 15% faster. So these are kind of the metrics that we see. We kind of roll it up all and say that, hey, what is the revenue per seller that you have right now? We will improve your revenue per seller by 20%. So that's kind of the metric that we try to drive our customers to. So that's kind of the metric that we try to drive our customers to, because this 2025 is the year where everybody is trying to do more with less and every CRO is thinking how do I go from my quota of 1.2 million to 1.4 million without adding additional headcount? So that is what we are trying to give to our customers, and our customers include ZoomInfo, demandbase, wordfix, pathify, newstore.
Speaker 2:The best thing that has happened is our AI has been in enterprise production for over 15 months, so that, again, if you look at the AI companies, there aren't enough AI companies with enterprise-level production deployment. For 15 months that has been going. We have got a few thousand users on our platform. Now. We have answered over 55,000 questions so far. In the last you know, 10 months or 12 months, let's say Wow, and almost all of our initial customers who went through the first cycle. They bought more. So they started at like 35 seats, then went to 230 seats, 500 seats. At ZoomInfo we support about 1300 sellers and a bunch of other places where they went 40 to 82 seats, things like that. So so far, so good. It's still early days for us and AI, but we feel very confident. That's amazing.
Speaker 1:Let's talk trust. Ai tools, as you know, can sometimes go off the rails as hallucinations, other problems. How do you make sure information is accurate over time? And even with the salespeople, how can you keep them on the rails? Is that also part of the goal to make your sales team more reliable and efficient as well?
Speaker 2:make your sales team more reliable and efficient as well. 100% the hallucination. If you look at the last two years of evolution of AI and LLM models, the conversations around hallucinations have come down Our customers. When we started the company, like one and a half years back, they would come to us and say, hey, what about hallucination? And we did this huge test where we proved to them that our hallucination was 2.3%, which was you know the industry best, and things like that. That was some time back. But these days, in the last five to six months, if I look at all the sales calls that I've been doing, the conversations around hallucinations are not happening as much. Honorizations around hallucinations are not happening as much because I think we are at a point where in the enterprise stack, companies have figured out multiple techniques to keep the hallucination very minimal. So if you take our system, 100% of our answers are grounded in the enterprise knowledge and 100% of the answers will have references associated with it. So that also ties a little bit to your question of how do you get the salesperson to trust the product and how do you make sure that the salesperson is using the product in the right way.
Speaker 2:One of the things that we do is giving references for every answers that we give, and these references are clickable click-throughs. You can go to the source and verify if you want to choose. The second thing that we have done with our customers is, whenever we give an answer, we use the recency of the data, frequency of the number of times we have seen the fact, authority of the source and authority of the person to give the answer. Then we package all of this into a confidence score, so every answer will have a confidence score that goes from low, medium, high, verified. So those four levels the salespeople will be able to see A bulk of the answers will be in high, but if it is low, then they know that they got to check out the references.
Speaker 2:The other thing that we have also done is we have added something called as a guard note. It's kind of like guardrail. So there might be topics that the companies are like hey, if it is about privacy, don't let the account executive copy paste the answer. So if it is about gdpr or privacy or pii or things like like, let's say, the uh, the, the compliance related topics, then our product has the ability to show up a note that says this talks about GDPR, you have to talk to the subject matter expert and this is the right person to talk. So there are so many pieces of workflows that we have implemented. This is just. I gave an example of GotNote, but we have implemented so many pieces. That gives the trust to the user and to our customers.
Speaker 1:That's great. Yeah, when I started in sales, I had pretty much a Rolodex, you know that whirly thing that circle that moves around, and a phone and a voicemail box and a pager and that was pretty much my tech stack. I mean, these days, of course, you have Zoom and Slack and Teams and Salesforce. How does integration work with the tools that your customers already use? How quickly can you get deployed?
Speaker 2:It's fairly straightforward. In fact, I am the strong believer that LLMs have made enterprise deployments way faster than it used to. I mean you and I. I know the enterprise deployments that used to take a year, one and a half years to get to some form of a pilot. There are still tools that are implemented that way. One and a half years to get to some form of a pilot there are still tools that are implemented that way.
Speaker 2:But by and large, if you talk to the AI vendors native AI vendors, not the vendors who had something. And then now moving to AI, if you talk to AI native vendors, one thing that you would start seeing as a common denominator is the onboarding is way faster because there is a lot of turnkey things that you can activate. To get to, let's say, time to the first value can be as early as hours to days to weeks, not months. If it is months, there is something wrong. At least to get to the first value. Then you can obviously deepen over a period of time. In our case, we typically see that we can get somebody to the first value in less than a week and the full deployment is about two weeks to three weeks for a midsize company and six to seven weeks for an enterprise company, that's because of the seamless integration right.
Speaker 2:We have 100 plus app integrations. So almost all the enterprise solutions that you are using, whether it is Slack or Teams or SharePoint, google Drive, intercom, zendesk, chorus, gong, document 360, seismic Highspot anything that you're using we can bring it in fairly quickly. There's no labeling, there is no tagging, because some of those complicated problems AI is taking care of organization of the data. That makes super easy for the admin. That is the reason why the onboarding is way faster, and I believe that other vendors have also gotten this advantage.
Speaker 1:Fantastic. So what's next? I mean you've raised money recently. You certainly know how to grow and scale companies. What's on your roadmap for the rest of the year?
Speaker 2:The focus today is product and go-to-market. Yes, you're right, we have raised two rounds. The first one was a 5.3 million seed round led by Foundation Capital and a bunch of amazing angels, and the second round was a Series A led by Mayfield and followed on by Foundation. We have enough money in the bank, so it's always good to have a war chest when the world is moving this fast. We have a team of about 40 people, so it's a well-rounded team. We have the right team to execute. We are not trying to find that one more talent to help us get somewhere.
Speaker 2:The focus today is product and go-to-market. The product has to move significantly fast. You may have seen the recent launches from OpenAI. It just came into the enterprise space so quickly, right, one fine day, one hour, one tweet. They are in the enterprise and into enterprise sales, into enterprise marketing. So if you are in the B2B space, you have to be really on your toes. So every day I spend a significant amount of time thinking about the product when should we go? How do we deepen it? How do we not become a roadkill by open AI? You know all of that. So think go-to-market. Obviously, go-to-market has not been this hard in the last 15 years. I can say Before that I do not know, because I came into the workforce 14, 15 years back.
Speaker 2:I've done two companies. I've sold other products and services before and after. Actually, it has never been this difficult to catch somebody's attention. The attention is a very, very, very high premium today and the noise is very high. So I mean, you know this, evan. You interact and interview so many amazing revenue leaders and founders and others go to all of the sales tech companies website. Everybody has the same tagline right, make your best, make your make every seller into a best seller. I can come, I can tell you 10 companies with that same tagline. So now I'm pretty deep in sales tech, right? I've been here 13 years and I go to different websites and I'm like huh, so this is the exact same thing that they are also doing. What are these guys doing different? If I am not able to understand that's playing in on the field, I cannot imagine a revenue leader to understand okay, this is what x is doing and this is what y is doing. The challenge with that is when you are super confused. The companies tend to take like a wait and watch mode.
Speaker 2:I don't know what this is. So let me wait a little bit for this noise to clear so that I see in the AI space. But on the flip side, the good revenue leaders, they also know that they cannot sit and wait because in the next two years or three years, max AI will give you a competitive advantage and in three years it will get to be table stakes. At that point if you are not full on in ai, then you are going back. So our job is to thread the needle in between this, get the attention, get through the noise and convince the people. This is the right thing to act. Um, we have been fortunate to be the Gardner Cold vendor and a bunch of other awards, so able to raise above the noise a little bit. But this is the typical world that vendors are operating in today, and my world is not significantly different. So product and go to market.
Speaker 1:Well, that's a very, very insightful answer. Let's move into a bit of the science fiction version of the discussion answer. Let's move into a bit of the science fiction version of the discussion. We're racing towards AGI. Agentic AI is really starting to take off. Do you foresee, you know fully virtualized sales agents who are sellers themselves? You know sort of avatar influencer types who can engage with customers. I mean, you're seeing bits and pieces with conversational AI, but you know, you think this will be a practical approach to selling over the next five years.
Speaker 2:If you go to docketio today, now our AI seller is going to pop up in the middle, bottom, middle, and it'll actually have a human-like conversation with you it'll talk in voice, it'll talk in text, it'll pull up slides, it'll pull up videos and kind of take you through the whole thing, capture your email, book your meeting, the whole thing. So so I guess my answer is yes. I do believe that AI will do a lot of sales conversations and will be the seller. But there are two ways to think about AI, evan, and this is very interesting. There are places where we are thinking about displacement of humans, right Like will an AI sell ServiceNow in the next five years? No, it is not going to be able to sell or service ServiceNow kind of a tool in the next five years. No, it is not going to be able to sell or ServiceNow kind of a tool in the next five years. It's not possible. Five is very difficult to say. Just to be clear, let's say three, because AI is so fast.
Speaker 2:So, let me pull that back and say three. But there are cases. Imagine you are going to buy an insurance, or imagine that you are going to buy a vacuum cleaner. The problem with that is if you go to, let's say, dyson or any of the vacuum cleaners, they have 50 vacuum cleaners or 150 vacuum cleaners. Now you are like I have no idea what to do with this, but you don't want to go to Best Buy. Right, best Buy, they will kind of guide you at least a little bit. You don't want to go to Best Buy, what do you do?
Speaker 2:So there are a lot of these cases where you need, ideally, you would love to have a human support, but for dyson to put a human to speak with all of you coming onto their website impossible, it's not. So there, and yeah, I can do a lot of. All AI has to do is three questions, right? So do you have pets? If you have pets, then obviously a bunch of account cleaners are gone. A pet friend leaves. So the 60 just came down to 15. Then the next question is is it a commercial use or a personal use? Boom, the 15 came down to three.
Speaker 2:Right Now, as a human, you can make the call Can an AI do this Absolutely, and you don't need AGI, you don't need ASI. Current systems can easily do it with very minimal intelligence, honestly. So there are all these use cases where the TAM today does not even exist because you are not deploying a human to do this. There is where AI is going to play when the ACV is low, when the complications is not that high, in other words, when the sales friction is high but the post sales is not that high. That's where AI will play a lot and it'll keep deepening itself a little bit, but I don't think that the enterprise motion that you and I know is not going anywhere anytime soon.
Speaker 1:That's so well said. Well, it's good news. Humans will be at the center, at least for the foreseeable future. Thanks so much for joining Really intriguing and exciting path you're on. Thanks for sharing the peek behind the curtain.
Speaker 2:Thanks a lot, Evan. Thanks for having me, and you enjoy the rest of the day.
Speaker 1:I will do. Thanks lot, evan. Thanks for having me and you enjoy the rest of the day I will do. Thanks everyone for listening and watching and be sure to check out our new TV show, techimpacttv, now nationwide on Bloomberg and Fox Business. Take care everyone, thanks.