00:00:06 - Bayram:
Okay. You should see my screen now. So if you have any questions, you just use the chat to ask your questions. Now, just a moment. I'll unmute you. Hold on.
00:00:34 - Bayram:
There you go. So, hi. So today we're actually kicking off our series of online events dedicated to AI in action. And in this season, we'll be talking about B2B sales and how you can scale B2B sales with automation and intelligence.00:01:03 - Bayram:
And today we will first discuss why this is relevant right now and what the market tells us about AI in sales and the benefits of applying intelligence to AI to sales processes. Then we'll continue building on that. We'll just take a look at the Gartner's 2024 report about the 13 generative AI cases.00:01:36 - Bayram:
For sales. Then I'll present our vision, how you can call it like AI Sales maturity model or five levels of autonomy of sales org, where basically this is how we see companies progressing from no AI use at all to autonomous sales orgs. And where we are right now and what we see coming up.00:02:06 - Bayram:
In future, then the framework that we use with our customers included to basically pick the quick wins that deliver more or less immediate results, and then think of the other use cases that could be applied and evaluated in a given sales organization. And then we have Abhinav and we have.00:02:36 - Bayram:
We will learn about alma's experience of applying AI in their vision overall. That's the agenda will take us about 60 minutes, so stay tuned. So first of all, I want to Gartner did a survey of chief revenue officers across different segments of companies and revealed a couple of stats that got me interested last year, and I wanted to share.00:03:07 - Bayram:
Share those with you? Well, first of all is that they predict that by 20, 26, 50% of time that account executives spent on prospecting and preparing for a meeting, those will be reduced. So we'll slash those. And we see that already that big chunk of preparation time in prospecting is actually can be.00:03:38 - Bayram:
You know, reliably executed by AI Agents. And these could be the immediate quick wins that you can capitalize. And that you can use to promote the use of AI In a sales org. Second aspect of that survey is that about a third of the chief revenue officers. Agreed that spinning up a separate generative AI operations.00:04:08 - Bayram:
Teams as part of their organization is something that they are planning this year because the survey was last year. And I actually see that some of our customers are basically combining the knowledge and experience of account executives with the capabilities and technical excellence of Genai engineers to push the AI and apply AI in different aspects.00:04:39 - Bayram:
Of their organizations, and I'll share some of those use cases today. Last but not least, this is a very controversial actually view that 60% of the workflows will be done through conversational UIs by 2020. In fact, some minor aspect, or, I don't know, maybe not so minor aspect of sales outreach process at ALMA.00:05:09 - Bayram:
Is done through Slack, and we see that some of our customers prefer messengers like Slack Teams or Telegram to control and monitor the results and the actions of the AI agents. So this is definitely happening in some way, but we envision much more than that. And Gartner Survey actually.00:05:40 - Bayram:
Confirms that. The other thing that they revealed is these 13 generative UI use cases for B2B sales, and they basically split them into three categories by the feasibility and the value. As you can see here, different use cases are in different quadrants of this graph here, for instance.00:06:11 - Bayram:
Some of the quick and likely wins could be value message creation. I'm sure some of you already use ChatGPT and similar services to help you craft the cold outreach message or to fill some part of your RFP response. Or, for instance, prepare meeting notes and action items based on the recorded call that you had. And I see a couple of assistants.00:06:42 - Bayram:
Joined AI assistants that joined our meeting now by Cyber in our own product that basically record this meeting. And in fact, in our case, for instance, they provide some real time information for me about all of the stuff that's going on on this call right now. And it probably asked for your email as well by now, and that will be used to implement some sequences that.00:07:13 - Bayram:
You would expect after a seminar or a webinar like this. But you can see that some of the use cases are, in terms of the value, they are pretty important. But maybe the feasibility is not there. But frankly, this is 2024. I see that now in April 2025. Some of the things like autonomous prospecting is actually a quick win. And that's something that.00:07:44 - Bayram:
I will share more. So when we think about AI sales or AI in sales orgs, we like to use the same metaphor or the same framework as they use for self driving cars. So basically different levels of autonomy depending on what percentage or what kind of what fraction of tasks are executed by AI versus a human.00:08:14 - Bayram:
And we think that given what's possible, right now we're somewhere between the L2 and L3 levels, 2 and 3. And I'll explain each of the levels and give you an example of the tasks for them. The zero level is actually no autonomy. Actually, basically everything is done by human. The 100% judgment and execution is done. And basically any task that you can think of in the B2B sales order.00:08:46 - Bayram:
And I'd say probably in 2024, most of the companies were there, but now I see that many of them actually, especially startups, progressing to at least level one, but most level two. So let's review what level one of autonomy is about. Well, at this stage, it's an assistive kind of autonomy, meaning that.00:09:16 - Bayram:
AI does some small tasks of bigger workflows and helps human to basically summarize to augment human in terms of their capabilities. For instance, suggest the next best action or summarizes the call or helps you personalize an email or LinkedIn copy. And this is very already this is useful.00:09:46 - Bayram:
And I'm sure many of you actually leverage ChatGPT and similar services for these purposes. In fact, many research proves that many employees are actually using ChatGPT in their job, but they never reveal it to their managers. And I think that's. That's the case with many of us, including me. So at this stage, it's more of an assistant job. We have human and human judgment on every. Each and every stage of the artwork.00:10:19 - Bayram:
Workflow and essentially think of this as an intern that prepares some work for you, but never actually takes responsibility for that work. Most of the software is in the space, but we see that AI agents are getting into the space as well. And again, the agents that joined this meeting and that will summarize this meeting and suggest some action items is the best example of this kind of level of autonomy.00:10:50 - Bayram:
The next one is partial autonomy, meaning that there is some process or workflow, say in an outreach or a post meeting processing and that workflow, some tasks or some steps of that workflow are executed by AI agent, but they require a sign off from a human. So what human does is.00:11:21 - Bayram:
Actually approves those results before agents change the state of our systems or change the state of the environment. Meaning that before they send a follow up to a customer, an account executive would review the meeting notes. Or before posting the results of the call to a CRM, the account executive would review and maybe tweak some aspects of.00:11:51 - Bayram:
The call and some values that are locked to the CRMs. But still this saves a lot of time, post meeting and follow up sending. And I think this is already very useful. In fact some of our customers, the very first use case that we started with was to lock the meeting notes, log the calls.00:12:21 - Bayram:
To Salesforce and extract some valuable information from those calls and push it through via Slack to the respective team, say a marketing team if any competitors were mentioned or some information was shared about the source of this prospect. Or for instance a product team if some use case or pain point was mentioned that is not addressed by the current product. Moving on to the next level.00:12:52 - Bayram:
Is a conditional level of autonomy, which is basically now, instead of just some steps of the workflow, the entire workflow is executed. But there are some, obviously, in the real life, there are always edge cases, there are always exceptions, there are always things that go not as planned. That's why on this stage, we need a way the role of us, of human.00:13:22 - Bayram:
Is Human is to monitor what agent is doing, but we will be notified if there is some edge case or exception. And human is to make a judgment around whether to override that exception and things like that. With Alma, we would, for instance, this is somewhere in between two and three. We would suggest a response.00:13:52 - Bayram:
To lead that we reached out to through our automated prospecting and targeting agent. But Human is in control in terms of actually overriding the exceptions and they can edit the messages that they are not fine with. Booking meetings, qualifying inbound leads, real time call coaching. This is something that is done through.00:14:22 - Bayram:
That kind of agents. Moving on to level four, by the way, again, most of us are in somewhere in between 2 and 3. So think of these as vision. Although some elements of the upper two levels, the fourth and fifth level of this autonomy, I see some elements of them implemented and deployed. So essentially the whole workflow is.00:14:52 - Bayram:
Implement autonomously, but there are some strict guard lines in place to basically reduce the number of exceptions. And obviously this is an evolution when typically what happens is that the agent starts to help or assist, then it does more of that about humanism, control human is in the loop. Then at some point the reliability of LLMs in.00:15:22 - Bayram:
Agent and the knowledge, the context that they have to respond to some requests and things like that is so good that organization decides to actually make it autonomous, but maybe in rare exceptions, handle those exceptions. So an autonomous outreach sequence is something that we do for many of our custom.00:15:53 - Bayram:
Customers. In fact, this is our key probably value prop. But I see that there are some other things like deal risk scoring done by some of the companies out there. And last but not least or probably is the most kind of end state is where the entire sales org is autonomously managed by AI. This is an end to end revenue engine that learns, optimizes and executes the sales.00:16:24 - Bayram:
Process and the human role is to basically set the objectives and guardrails. In fact, I think human's role in any AI, AI assisted or AI executed process, the human role will shift to actually controlling setting objectives and controlling setting guardrails to AI. And of course, dynamic pricing is probably one of the.00:16:56 - Bayram:
Examples. But what I see that could be done, and we're very close with some of our customers, is when our autonomous prospecting workflow works. But the AI spots some new micro segments of customers that are not part of the initial targeting by our customer, but in fact that micro segment converts.00:17:26 - Bayram:
Better. And that's why we suggest to expand the targeting. For instance, with one of our customers, just recently, we spotted that people that are part of the Forbes 30 under 30, those people are the great prospects in terms of converting to a customer. So we suggested expanding targeting to include that segment because that was not.00:17:56 - Bayram:
Part of the initial targeting. So what I mean here is that where AI would spot some micro segments and suggest to expand the targeting, expand the total addressable market for a company. And again, I see elements of that happening right now. And the way we think about rolling out any AI strategy in the sales work is to basically what we call.00:18:27 - Bayram:
Explore versus exploit. So basically most of the stuff that you want to do is exploiting the industry best practices, having those quick wins to show the promise of AI and show quick results, thereby gain trust and resources to implement some other aspects. And then after quick wins you get to controlled experiments and a moonshot idea. I'll give you an example.00:18:58 - Bayram:
Of a rollout, and that will give you an idea how this could be rolled out in your organization. So the quick wins. These things are proven things that save time or increase conversions, and they can be rolled out pretty easily. So Meeting Preparation report recall the 50% number on the first slide that.00:19:27 - Bayram:
The reduce in the meeting preparation times. Actually one of our customer we process thousands of prospects, we help the account executive to prepare for the meeting and we would basically AI agent just learns and sources proprietary and publicly available data by the prospect and prepares a one page brief that gets sent to.00:19:58 - Bayram:
30 minutes before the meeting, and that significantly cuts the time an account executive needs to be prepared for the meeting. So that's a very quick win that I would strongly encourage everyone to apply. Second, quick win is logging cold nodes to CRM and some note takers like Cyber here. They have native integration with the CRMs to implement.00:20:28 - Bayram:
Implement just that. You can go further than that. You can extract some values from those transcripts. Like for instance, applying a medic framework to extract some important signals and push that information straight to the CRM. Saving 50% of the time account executive would spend on basically logging the results. So no manual work to log the results and you are happy.00:20:59 - Bayram:
Because first of all, data is in the CRM means it improves the revenue intelligence and revenue forecast features of the CRM. But at the same time, AES are happy because I don't know any AE that is happy to lock the results of the calls to the CRM. What's interesting is that sometimes we notice with some of our customers that AES sometimes are too optimistic about their calls.00:21:30 - Bayram:
And when we compare the results or the way they would log a given call to the CRM with an actual transcript, it seems like sometimes they're too optimistic. And I think many AES are optimistic initiative, and that's great. But sometimes that just makes our forecasts of less quality, which is something that we want to avoid. That's why logging call notes is.00:22:00 - Bayram:
A very quick win that you can apply. And last but not least of the quick wins is to basically qualify inbound leads by enriching sourcing information available online and in some proprietary database and triggering some workflows or sequences of actions like, for instance, requesting additional information or asking for assigning this high.00:22:31 - Bayram:
High probability inbound lead to sales rep. These sequences and these workflows could be triggered automatically based on the lead qualification. What we realized that sometimes AI agents do a much better job of these kind of things. Like for instance, in one case, our AI agent pulled information from publicly available government databases that we.00:23:01 - Bayram:
We and AES had no idea about to source some signals to basically qualify this lead. And the Reasoning models like O3, for instance, do a great job in terms of designing a plan how to source that information. And then you would use some tools to actually get access to that information and put that information into into the context of an AI agent to basically increase.00:23:31 - Bayram:
This ratio of how many sales qualified leads you get per prospect and we see gains there. There are a couple of, I'd say less quicker or longer experiments or longer bets that you can do that I know work, but they require more preparation, they require more configuration, and they require.00:24:01 - Bayram:
Some back office changes to implement, but in fact the automated outreach where AI agent actually identifies prospects, drafts the messages, monitors engagements, sends follow ups, and automatically books the meetings on A's calendar. This is something that works, for instance, for Alma. Not 100% of it, but.00:24:31 - Bayram:
I think about 75 80. We'll discuss that with Abhinav a little bit later. But in fact this is possible and you can implement an automated outreach and basically help AES to focus on closing deals rather than sourcing the prospects. And in bigger companies, of course you could have a dedicated teams of SDRs doing this job, but in a smaller companies or companies that want to.00:25:02 - Bayram:
Stay lean and AI agent assisted automated outreach is something that you can implement from a level 3 autonomy to reduce the load on AES generating leads. The second one is life coaching and debrief where during the call as you can see on CI sales associate joined our call and actually in real time it provides me.00:25:32 - Bayram:
A transcript of everything I say. I can ask a question or instruct it to suggest me some questions. If that's a customer discovery or customer interview call after the meeting. Obviously it would grade the call, prepare the meeting notes, and suggest the next best action to take, which eventually increases the meetings to deal.00:26:03 - Bayram:
Conversion rates and the moonshot that I've mentioned on the previous slide moonshot bet is to basically mine the win loss data from CRM and monitor some external signals. Like for instance, a champion from one company joins another company and you can reach out to them to close that customer. This is something that could be automated.00:26:33 - Bayram:
Autonomously. But again, this is a moonshot. I see some elements of this happening, but we're not quite there yet, even though some aspects of this could be implemented right now. So that's in the nutshell why we think AI could be applied in sales orgs and should be applied. What could be the strategy of, you know, applying AI? What are the quick wins and some moonshot ideas that you could.00:27:05 - Bayram:
Apply. And let's transition to our discussion with Abhinav, chief of staff at Alma. And I will just. Hold on. I'll just switch to and let you unmute. Abhinav, you should be live00:27:29 - Abhinav Kumar:
now.00:27:30 - Bayram:
Yeah. That's great. Hi, Abhinav. And thanks for joining us.00:27:37 - Bayram:
Today. So I will ask a couple of questions first about you, and then your experience with the ONSE and overall, your experience of applying AI in your organization and sales organization specifically. And then we'll end up with what's your overall vision towards the benefits and promise of AI in sales? Okay, so since not.00:28:08 - Bayram:
Some of us may not be familiar with Alma. Could you please tell us more about Alma? And what are you guys doing?00:28:16 - Abhinav Kumar:
First of all, thank you so much, Bayram, for having me here. Excellent presentation. So, a quick note about Alma. We are an immigration legal tech platform. What we are essentially building is the future of reimagining immigration law. Think about immigration law firms like Fragaman and bl. What you want to do is do similar revenues, but with a fraction of the headquart.00:28:37 - Abhinav Kumar:
Account of attorneys and paralegals. So when Bayram spoke about like the five levels of autonomy, we're also thinking about the same thing, but from an immigration legal perspective. So what are the different tasks that can, you can keep automating over time and make the attorneys like 10x more efficient. So that's what we're doing. We're working with all sorts of employment based visas. So you work with founders, researchers, early stage employees to get their O1s, EB1 as EB2NIWS.00:29:07 - Abhinav Kumar:
Hnbs, tns, all sorts of employment based visas and dependent visas. Yeah, I mean if you guys have any immigration needs, happy to do a consultation after this call.00:29:22 - Bayram:
That's great. Thanks Abhinav. I actually know I think at least three of my friends founders who are building AI companies thanks to Alma and you and your.00:29:37 - Bayram:
Your colleagues actually have their own S approved. So thanks for that. I think this is a great service that you guys are offering with a great accuracy and turnaround times. So, chief of staff, what is chief of staff? Tell us more about that and how you got to the point of being a chief of staff.00:29:58 - Abhinav Kumar:
So, yeah, that's an interesting transition. So just a quick background about me. I'm from India, so I used to I graduated my undergrad in 2017, worked with.00:30:09 - Abhinav Kumar:
Bain & Co. Out of India for five years as a management consultant. Came to the US to do my MBA. Did that in New York. And then yeah, I was supposed to go back to Bain and Company after my mba, but they pushed out my joining. I wanted to, you know, take a year and like experiment with a pre seed company because ultimately I do want to start my own company down the in the future, like maybe a few years from now. And I thought, like, what is the best way to, you know,00:30:40 - Abhinav Kumar:
Get my hands dirty. Essentially just join a pre seed company. And so yeah, reached out to Azada, who was the founder and CEO of Alma. They were like three people at the time. Created my own role. So the chief of staff was not the role that existed then in the company. Like they're not recruiting for it. I reached out to her, sold myself, and then, yeah, created this role. So I've been wearing all sorts of different hats here. So I think it's for anyone who wants to become a founder and is not yet ready to take that plunge. I feel like that chief of staff role is.00:31:10 - Abhinav Kumar:
The perfect role to do that, because you do everything that a founder does, obviously without title and all. But, like, you get to. Like, I am right now, I'm recruiting, I am leading gtm. I am handling the finance and accounting of the company. I am, yeah. Like, I'm setting up operations across the board, so doing a lot of operational stuff as well. So it's been. It's been very interesting. Highly recommend this role for whoever wants to become a founder in the future.00:31:39 - Bayram:
Yeah, yeah, definitely. Like.00:31:41 - Bayram:
Very multifunctional cross00:31:43 - Abhinav Kumar:
discipline.00:31:44 - Bayram:
Yeah. And you can learn a lot, I think, in this capacity. That's great. You mentioned that one of the roles is basically a go to market leader. And you focus on this function and try to recall, I think late last year when we met each other and you decided to trial Onsi AI at Alma. Could you tell us more about.00:32:13 - Bayram:
What was the key metric for you? How, how did you approach a trial and what was the, the definition of success for you in that trial?00:32:24 - Abhinav Kumar:
Yeah, I, I remember when we first spoke Bayrams, I think we were doing a one or two month pilot where we were just doing a couple of thousands of reach outs from multi, modal, multichannel reach outs from LinkedIn and email and then.00:32:43 - Abhinav Kumar:
That one or two month period was essentially to understand whether you want to get into a longer term engagement. And the metric that I was looking at back then is essentially like dollar value of revenue that we are able to get out of this pilot. Right. So I think if I remember the numbers correctly, of the 2,000 or $3,000 spent, we got like 50, 60, $70,000 in revenue out of it. Right. Which is like you don't get that kind of return out of every channel. So that was like the.00:33:13 - Abhinav Kumar:
Metric that, like, you know, really pushed us to double down onto this relationship. So thank you for that. Thank you for reaching out and helping us, like, build one of our most important channels, which is outbound. And not just that, but also, like, adding more and more features across entire sales ops journey. And then I think, Meera, when you're talking about, like, the different levels of autonomy, I was just thinking through, like, what are the. Where are we at different parts of the funnel? And I think we are at different levels in different parts of the funnel.00:33:43 - Abhinav Kumar:
But the goal for this year is essentially build on this relationship and, like, sort of add more and more autonomy and, like, basically add more and more leverage on our. On the GTM ops time and also the account executive's name. So, yeah, I mean, like, we've seen a lot of results, and that's why we continue investing in this.00:34:04 - Bayram:
That's great. Speaking of the other aspects of the funnel and other stages of the funnel, obviously at some point,00:34:13 - Bayram:
Point you decided to apply AI to the inbound lead qualification. Why this decision was made and what were your expectations out of this? Why do you think AI is relevant and useful there and how you envision AI assisted lead qualification? What are the results for you as00:34:39 - Abhinav Kumar:
a company? So, just to give some context, background to the audience here, so.00:34:45 - Abhinav Kumar:
We are using Onza for outbound, and both like lead qualifications. So on outbound, what happens is that MERAM has helped us build algorithms which automatically finds ICPs, people that fit into our ICP, and then automatically qualifying them and reaching out to them and then sending out the relevant messages and whatnot. And so we thought that if the algorithm is already built to qualify the people for outbound, why not leverage it for inbound as well? And the rationale there is that.00:35:15 - Abhinav Kumar:
Getting a lot of inbound as well because of other channels that are working. Right. But on the inbound, like, it takes a lot of time for the account executives to go through a profile. So in immigration, for example, you have to like, as an account executive, you have to look at everything that is available online to understand whether a person is eligible for a particular visa or not. Right. For example, if you were to talk about the O1 visa, which is an extraordinary ability visa, you have to understand whether the person has won a review award. Is the person a member of, like, certain associations? Do they have test coverage around them? What is their Google Scholar score?00:35:46 - Abhinav Kumar:
Hindex score. So there's so many data points that are available either through the resume, the LinkedIn, the Google Scholar there, you know, if they're a founder, they like data on Crunchbase and pitchbook and like all sorts of publicly available data. So an account executive, without having any sort of an assisted lead qualification mechanism, spends roughly like 10, 15 minutes per inbound lead, if you are able to scale their time by just doing that work. Because, for example, imagine that a lead is getting like,00:36:16 - Abhinav Kumar:
And the account executive is getting like 20 leads a day, right? That's easily like 200 minutes or 300 minutes spent on only lead qualification, which is just reduce that, cut that by a little, but like maybe two, two and a half hours on this, like, qualifying, really qualifying certain leads. But if you get some sort of an AI assist, which is essentially doing that work for you, that literally cuts the time into 10 minutes versus spending three hours on it. So that is what led us to sort of build that, you know, collaborative.00:36:47 - Abhinav Kumar:
Build that lead qualification score and it's working pretty well. So basically the. But having said that, the account executives still use it as don't use it as a substitute, but as a complement to their own analysis. But it's certainly shortcut the time by, I'd say like 70%, which is a huge time save. That opens up time for just doing more calls, essentially.00:37:08 - Bayram:
Right? Yeah. Yeah.00:37:09 - Abhinav Kumar:
This is a more productive use of their time.00:37:11 - Bayram:
That's right. I remember that your founder, Aizada, was.00:37:17 - Bayram:
Sharing this that there was some prominent investor that tweeted about ALMA on Twitter and that generated a lot of inbound leads in a very short time. And in addition to being able to save ease time on qualification, this is about the faster, I think reaction because by qualifying automatically inbound leads on Saturday evening, you can prioritize.00:37:48 - Bayram:
Those leads that should be processed faster and have the conversation with ACE earlier than later. And that helps you to close tasks, I00:37:59 - Abhinav Kumar:
think. But just to add on that, I think we're still at level two or level three there, but the goal there is to move to the next level, essentially, where we refine the lead qualification to an extent where it does better than an account executive doing it manually, which means that then we can build an automation that if someone has received a high score, we just automatically send out.00:38:18 - Abhinav Kumar:
The call and write versus waiting for an account executive to review the lead qualification. So that's where you want to ultimately move in the next few months. Which like sort of because. Which helps in improving the experience for the client ultimately. Because over the weekend, if you get certain leads, they don't have to wait until Monday to be to hear back from.00:38:37 - Bayram:
So. Yeah, that's right. Makes sense. So closing the second like part of our discussion here, what is the net new capability that you expect.00:38:49 - Bayram:
From AI agents, not necessarily on AI, but GTM AI agents. That would move the needle the most for alma.00:38:59 - Abhinav Kumar:
I think for alma, we are doing a lot of automation. Top of funnels, we're doing a lot of outreach. We're doing a lot of outreach. And then messages and then booking automated calls, we're doing that. Middle of the funnel, we're doing lead qualification. I think the scope is after getting on a call or once you get on a call, I think the net new capability that's going to help.00:39:19 - Abhinav Kumar:
Move the needle is essentially enabling the sales team to do a better job on the call. So essentially being able to sort of guide them. What are the best next questions to00:39:30 - Bayram:
ask?00:39:31 - Abhinav Kumar:
And how do you get a lead, a qualified lead to closure? Right. How do you improve conversion after you got on a call, especially in a high volume and high velocity environment, Sales environment. So I think that I would say that is where we see, like,00:39:50 - Abhinav Kumar:
The next new big feature being added00:39:53 - Bayram:
into00:39:53 - Abhinav Kumar:
our GTM operations.00:39:55 - Bayram:
Makes00:39:55 - Abhinav Kumar:
sense. And I think you spoke about this, touched upon this essentially having some sort of agent which on the call guide the account executive real time and it's being trained on all the closed one calls historically on what really went well and what are the things you should be doing versus not. And feeds on top of the immigration knowledge base and sort of guides the account executive to ask the.00:40:20 - Abhinav Kumar:
Right set of questions and tells them do this, do that, and then also guides them post the call. Right.00:40:26 - Bayram:
Yep. Makes sense. That's great. So moving on to the your vision about AI in sales works. So fast forward five years. Of course, if we are still around, you know, with all the AI fast take scenarios. But five years fast forward, what does how does.00:40:50 - Bayram:
Sales org look like? What are the jobs of AI versus the human? And what would you expect those to be and why?00:41:02 - Abhinav Kumar:
That's a very interesting question. So five years is a long time in AI, especially now. So I think even in the next one or two years, I feel like a lot of the best sales dogs are going to move to level three and a half, four,00:41:16 - Bayram:
and00:41:17 - Abhinav Kumar:
in the next five years, definitely I see a world where we move to level five.00:41:22 - Abhinav Kumar:
If I were to take the example of Alma, for example. In the next two years, you want to move to a place where. One, one and a half years, you want to move to a place where, like, the account executive only has to get on a call00:41:33 - Bayram:
and00:41:34 - Abhinav Kumar:
nothing else.00:41:35 - Bayram:
Yeah.00:41:35 - Abhinav Kumar:
Right. All the meetings are automatically booked. The lead qualification happens automatically. They're, like, entirely prepped to get on a call, even if they're not. Like, the best questions are available, like on the call, through an agent, and then everything after the call also happens automatic.00:41:52 - Abhinav Kumar:
Through, just pushing the call summaries as, like, custom properties into the CRM, creating deals out of, like, hot calls, and then just, you know, guiding the A through closure. But in the next five years, there's also, you know, a lot of these agents which are popping up, which are essentially going to replace AES, essentially, you know, be the human, you know, the agent ae, where it's all automated voice and.00:42:22 - Abhinav Kumar:
And video. But in our line of business, I don't know how relevant or, like, how safe would that be, because ultimately customers do want to talk to a real human while on a call. So if it's definitely, if it's like video modal, then at least in our business, I don't see that happening in the next one or two years where, like, our account executives are being replaced by, you know, a bot on a call. So, but maybe in the five years if, like, the methods get so advanced that you can not video but, like,00:42:53 - Abhinav Kumar:
Do a voice call and then sound human. But again, this is the question of you have to tell the other person. Right. Like, you have to clarify to the other person that you're talking to about00:43:02 - Bayram:
an00:43:02 - Abhinav Kumar:
arty human. So, like, those risks sort of exist. But I do see, like, the account executives getting maximum leverage on the time in the next one or two years by just doing the call and nothing else.00:43:11 - Bayram:
Yeah.00:43:12 - Abhinav Kumar:
Like, everything is automated and we're already seeing that happening across different parts, but it's all about, like, slowly just to keep adding another layer of automation on top.00:43:22 - Bayram:
That's right.00:43:23 - Bayram:
Yeah, that makes sense. And the cost of mistake is too high, like for a person that want to get legal, like oh,00:43:31 - Abhinav Kumar:
one.00:43:32 - Bayram:
And that's why we want to move slower than maybe in some other aspects of this job, because the cost of mistake is too, is00:43:42 - Abhinav Kumar:
too large. That's why I00:43:43 - Bayram:
feel00:43:43 - Abhinav Kumar:
like every new net capability that we add on the00:43:46 - Bayram:
automation00:43:47 - Abhinav Kumar:
side is essentially goes through a significant period of human in the loop experimental phase before we sort of.00:43:53 - Abhinav Kumar:
Put it on full auto.00:43:55 - Bayram:
Makes sense. So to wrap up this visionary part, are there any other technologies or capabilities, not necessarily necessarily of LLMs, but that you're excited about and that you envision, could change the way sales work broadly, like in different aspects of our life.00:44:24 - Abhinav Kumar:
Apart from LLMs, I do see, like, there are a lot of voice agents, voice companies that are popping up, which are doing, like, an amazing job. And for example, Cartesia, which is building, like, a foundational voice model, so which a lot of applications can use, which is, like, has amazing applications in healthcare and so on. But I feel like maybe not in immigration, but in a slightly lower risk business, I definitely see an environment where that can add.00:44:54 - Abhinav Kumar:
A lot more efficiency gains. Maybe you don't need account executives then,00:44:58 - Bayram:
right?00:44:58 - Abhinav Kumar:
You don't need00:45:00 - Bayram:
like00:45:00 - Abhinav Kumar:
especially in lower risk businesses, not immigration, not healthcare, not finance related. I do see like that being the next frontier of technology which can add 10x more leverage of the account executive team. Right. Like then all the account executive or like the person who's closing associate oversee the system of agents and make sure that everything's running smoothly and just making sure that there's no errors happen.00:45:25 - Abhinav Kumar:
Across, and then over time, that also gets replaced. And then it's just like, one person who's the engineer, like GTM engineer, who is sort of, like, doing everything end to end. Clay talks a lot about. Clay is this B2B outreach company which talks a lot about this new role popping up, which is called GDM Engineer. I feel that's going to become super relevant and, like, the roles of an account executive, sdr, bdr, you know, customer success, everything's going to be merging into, like, one, this one particular role.00:45:55 - Abhinav Kumar:
Which sort of works for different tools and automates the entire sales process end to end. But, like, bringing back to the question, I do feel like voice may be like the next big thing in a lot of businesses.00:46:10 - Bayram:
Yeah, makes perfect sense. So to wrap up our conversation, what is your recommended path to other GTM leaders? Head of.00:46:25 - Bayram:
Of sales? What is your recommended path in terms of evaluating the benefits and pitfalls of AI in their sales source? What's your kind of recommended path?00:46:39 - Abhinav Kumar:
So I'd say that in my experience in the last one year, as you set up the GTA motions for our business, especially on the individual side, I realized that the past to 10xing.00:46:56 - Abhinav Kumar:
Any. First you have to go from 0 to 1 on different channels,00:46:59 - Bayram:
and00:46:59 - Abhinav Kumar:
that happens manually. So00:47:00 - Bayram:
you have00:47:01 - Abhinav Kumar:
to experiment with what really works. You have to understand, like, what is the level of quality that you need across different channels to go from 0 to 1?00:47:09 - Bayram:
And then00:47:09 - Abhinav Kumar:
once you figure out, like, what channels really work and what channels don't, then you double down on, like, how do you go from 1 to 10? Right? That's where you start, like, automating stuff. And when you're automating different channels, I would say that always start with, like, a human in the loop, so.00:47:26 - Abhinav Kumar:
Don't go about automating that entire for example, if you take the example of lead qualification, as you said, right? Start with a lead qualification which is being read by an account executive and then you know the mail is sent out or like an invite is sent out versus just directly sending out an invite so that you know that things are being done accurately, even with lead outreach. Automated outbound, right? The messaging which is sent and the response. The messaging, like automated messages, is one thing, but the replies to.00:47:56 - Abhinav Kumar:
To a person who does respond, that should obviously be checked and then be sent. But so you do that for like a couple of months. See, like, if things are like the automated responses that are working well and then you fully00:48:09 - Bayram:
automate.00:48:11 - Abhinav Kumar:
But that's for like 1 to 10 state. For 0 to 1. I'd say do everything manually. Yeah. To figure out what really works. Because before you spend time in automating, which is like a huge commitment, you don't want to be automating the wrong things,00:48:25 - Bayram:
and00:48:25 - Abhinav Kumar:
you have to do it, like, the entire GTM experience.00:48:27 - Abhinav Kumar:
Experiment in a, in a mathematical way. You know, these are the 70 different channels. These are. You put like, different amounts of effort. The goal of each channel is to bring in SQLs that you ultimately close and you understand which one is bringing most SQL with minimum cost and effort. And those are the channels that ultimately work. Right. And that's when you sort of, sort of start automating.00:48:48 - Bayram:
Yeah, makes sense. Sounds good. This, this is great. I'll open the floor for a couple of questions. So if you have.00:48:58 - Bayram:
A question to ask. Either use chat or just raise a hand in Zoom and I'll let you unmute and ask your question. While we were sourcing the questions abhinav, maybe I mentioned that you use slack for monitoring and controlling what's going on with the AI agents. Can you tell us more?00:49:28 - Bayram:
What is the role of Slack for this use case? And do you believe that number that we had shared by gartner that by 2028 I think it is that many organizations will actually interact with sales data through conversational UI slacks and other messengers?00:49:52 - Abhinav Kumar:
100%. I mean 2028 definitely. I think that's going to happen sooner than later.00:50:00 - Abhinav Kumar:
So that's for sure. Just so your question is like, how are we using Slack in00:50:04 - Bayram:
our00:50:05 - Abhinav Kumar:
entire00:50:05 - Bayram:
day? Yeah,00:50:06 - Abhinav Kumar:
so we as I said, like, we use on the support and like automated finding ICPs, reaching out to them and whatnot. So we've also done vedram has helped us like with this integration where We've connected multiple LinkedIn accounts of all our team members and then the automated outreaches happen and whenever some responds.00:50:28 - Abhinav Kumar:
They actually show up in Slack. So you don't have to manage different LinkedIn accounts by logging into the accounts, but you can manage everything from Slack. You can literally approve messages or approve, like, the responses to the people who did respond to the outreach, or edit the message there and then, and then respond to the outreach. So that's an incredible save of time. Right? Like, you don't have to log into different accounts. You don't have to because, like, LinkedIn is also polluted with a lot of inbound, right? So you don't have to, you don't want to scroll with.00:50:59 - Abhinav Kumar:
All of the messages and find which one outbound ones and you have to respond to it helps you manage all the outbound in one go. And then now we are working. I think I believe we are working in automating that even further. But training the model on the edited responses ultimately, because I feel like I have been managing my own outbound at this point. So I have not been clicking on edit. I've been just sending the approve button00:51:23 - Bayram:
because00:51:24 - Abhinav Kumar:
I feel we are at a stage where it's very smooth.00:51:28 - Abhinav Kumar:
Yeah, so. And it's resulting in so much. Imagine not having to hire an SDR who does all of00:51:34 - Bayram:
this.00:51:35 - Abhinav Kumar:
That's so much cost saved in commissions and everything. And like also on time because you're able to do what an SDR is going to do manually. Just then it's it.00:51:46 - Bayram:
I00:51:47 - Abhinav Kumar:
do see. Vlad has a00:51:48 - Bayram:
question. Yeah, that's right. So, Abhinav, could you elaborate on how you evaluated various AI sales solutions and what specifically led you to choose onto.00:51:59 - Abhinav Kumar:
I think we did a pilot with a couple of these. I won't name the other solutions, but maybe you can chat offline. But I think the things that stood out for me essentially were just how hands on Veda was, just being able to work collaboratively in creating solutions that work for us was the key differentiator. And also in terms of the output, we definitely saw a clear difference in roi, a clear difference in.00:52:29 - Abhinav Kumar:
How at what scale we were able to reach out to different like our ICP and, and how, yeah, quickly and like basically reaching out to the precious people that fit into our ICP and like the warmest folks is what ANSA helped us with, with the other agencies. And there's so many tools out there. Right. And you don't get to chat with the founder across the00:52:53 - Bayram:
different00:52:54 - Abhinav Kumar:
companies. And like all of the self serve ones were not, were not as.00:52:59 - Abhinav Kumar:
Flexible a solution as onsite.00:53:04 - Bayram:
Okay. Sounds good, I think. Yep. Thanks, Vlad. I think that should be it. And we can wrap up here abhinav. Thanks a lot for your time and for your. For sharing your knowledge and experience. And obviously, we're learning from you as well here at onsa. And I like that you're constantly experimenting with other icp.00:53:31 - Bayram:
Piece in learning from the market to expand it and to service it better. So thank you.00:53:37 - Abhinav Kumar:
And then just like one last00:53:38 - Bayram:
point.00:53:39 - Abhinav Kumar:
If you're not using AI, you're not doing it right. Because I just feel like in today's world, AI in sales just helps you move 5x faster than your competitor. And not using is like a strategic, doesn't make sense strategically. So if you have any questions, if anyone has any questions about.00:54:01 - Abhinav Kumar:
Like how we are incorporating AI in different parts of the workflows, especially on the GTM side. Happy to chat about it. If you have any specific questions on ansa, Happy to chat about that as well.00:54:11 - Bayram:
That's great. Thanks. And actually Alma hosts some events for founders in terms of go to market and some other aspects. And if you decide to get a visa specific type of visa, please try Alma. I think this is a great service that.00:54:31 - Bayram:
Is confirmed by at least three of my friends Founders Abhinav, thank you so much and thank you everyone for joining this session. Yes, there is a link that Abhinav shared and I'll follow up that one as well in a blast in Alumablast after the event and I'll share the recording and the deck. Thank you so much guys and you have a good week.00:54:56 - Abhinav Kumar:
Thank you.00:54:57 - Bayram:
Bye.