About AI Agents for SaaS companies

Okay. I think we are live. Just turning off a couple of apps on my device so that I don't get a call or anything like that in the middle of the live. But, let's just wait. Okay, perfect. So we have our first few viewers who have joined in. So we know for a fact that we are live on YouTube and on LinkedIn. So we can start. Hello and welcome.

to Momentum Office Hours. My name is Yash, and I'm joined by my co-founders Kaushik and Jay to discuss topic of the week, AI agents for SaaS firms. Our goal is to provide you with actionable insights and practical strategies that you can apply to your own products. Throughout the session, we encourage you to engage with us by asking us questions and sharing your thoughts. This is a fantastic opportunity to learn from each other and give you insights that can help drive your SaaS initiatives forward. So let's get started. Jay, Kaushik.

How are we doing today? Nice. Good, good. Preparing for tomorrow's calls, work, everything. Preparing for tomorrow's call and work. Awesome. I am yet to prepare for my day tomorrow. I am going to be travelling. Going out for a small vacation over the weekend. So that's what I'll be doing. I don't think, I I've prepped for the last few months.

By working to be ready for the vacation. How are things with you Jay? Going great. Couple of things coming and going, setting few things up and upon the same. Awesome. So then let's get started. Let's talk about AI agents for your SaaS firm. What are AI agents? Why do need them? What is happening in this world? Yes. So before we jump into what AI agents for SaaS firms are, wanted to

have a fair understanding on what exactly is an AI agent and like how would it differ from the traditional AI systems that we have? So, primarily the difference between an AI agent and everything up until today I mean not today I mean let's say up until a year ago so all the technological all the technological advancements that has happened in the history is extremely different from

the AI agents that we have today. And the primary difference is that an AI agent is not a technology, is not a tool, but it in and of itself can perform actions. So what this means is take something that is extra. So let's go out of the digital world. So let's just talk about technology in general. The most powerful thing that we've developed physically also.

is like an atomic bomb or a hydrogen bomb or something like that. So hydrogen bomb is the most powerful thing that we've built. That is a technological advancement that we've done. Most people including myself would agree that it's not, there are more cons to that than pros. But that's the most powerful technological advancement that we have. It cannot perform actions by itself.

Doesn't matter however powerful it is, it is still a tool, it is still a technology. With AI agents, it can perform actions by itself. And that's the fundamental difference that you have between AI agents and everything else that happened before. So it's not like before JATGBT or OpenAI became popular, there was no usage of AI. So we had machine learning.

We had natural language processing. was used in our day-to-day lives as well for figuring out what are the ads to show to you on social media platforms. It was part of the video recommendation engine on YouTube. It was part of a lot of things that we were exposed to in our day-to-day lives. So there was AI for sure. However, before AI agents, everything that we had with AI was a tool or a technology. Once AI agents came into the picture,

they can actually perform tasks and they can actually do actions. That's the primary difference. Interesting. it's just we can say it's replacing human work, human tasks that people were actually doing. Now, AI would be able to perform that. Up to certain extent. So, one of the use cases that comes to my mind is the chatbot side, right? Where let's say a customer support chatbot is there. So,

especially now talking in terms of SaaS products or any other technology for instance where AI chatbots can be used. Are these AI agents different from AI chatbots that we see or is it the same thing, just a specific use case? So I would say that an AI agent is an extension of a chatbot. We had chatbots before agents as well. So a chatbot essentially would be able to

sort of acquire its own intelligence and give answers to the questions that you have. Give summaries to the long knowledge base or summaries to books or summaries of videos and movies and things like that. All of those things can be served through a chatbot. And so chatbots primarily were used to make sure that the human intervention happens for as little time as possible on a ticket.

so that all the basic details are collected by the chatbot. All of those things are happening through the chatbot. When AI agents come into the picture, they can actually resolve those tickets to a certain extent as well. So previously chatbots were only able to get contextual data around a ticket, around a problem. But AI agents can now, as an example, issue refunds, can now help you

you know, make payments can perform actions on your behalf and it can also perform actions on the actual support agents behalf as well. If you look largely at an industry, a good way to think about it is that if you look largely at the market or at the industry, so you have let's say 10 or 12 major industries, right? So you have technology, you have design, you have marketing, you have manufacturing, you have

You have healthcare, you have retail, you have all of those sort of largely distribution of the market or distribution of the industries that exist. Up until now, the tool or the technology that we had were replacing blue-collar workers. So if you were using AI as a tool, you will need less of blue-collars, you will need less of all of those things.

With AI agents, the more amount the job impact that it is going to create is going to be on white collar workers. So, who are the people who are sitting in offices with a device with an internet connection, who are getting on calls and who are doing designs and who are doing marketing and who are creating reports and who are updating metrics and all of those things. So, AI agents can actually create metrics for you. AI agents can actually create designs for you, can create videos for you.

can create images for you, can create graphic designs and things like that. So an AI agent is in that way significantly different than an AI tool or an AI technology. I think if one has to understand this for every SaaS product there is a software aspect and there is a people aspect where the AI agent can do both at the same time.

It sort of leads me to that question where is that what would you like to define as the best way for them to implement an AI agent in a SaaS product or you know to have that where can they provide the best value at an initial stage of a SaaS product with respect to an AI agent. So a good way to think about it is identify the workflows that are happening so if you're running a SaaS company

So let's say you have a SaaS company, have about a thousand users. For the purposes of this, let's assume it's a CRM. So let's say you build a CRM as a SaaS product and you have a thousand people who are using the product. Identify all the steps that 80 % of, let's say 800 people out of those thousand people do recurringly as a part of their workflow. Some of those things.

can be automated through just creating workflows, right? Creating the ability for you to automate things. as a SaaS company, when I build a workflow automation module as a part of my CRM, it's essentially an algorithm, right? Essentially a set of instructions. Now imagine that for all the other things that are repetitive in nature, but that need not just instructions, but need guidelines.

So in those cases, you can automate those using AI agents. So I'll give you an example for that. So an example for that is, let's say for this CRM, I'm a user and there are a thousand users like me. All of them get on calls with their prospects and then come into the CRM and then upload a meeting summary or write down a meeting notes and stuff like that.

and then take some amount of action, is where they will either move the prospect from a particular stage to another stage. Sometimes what might also happen is that when I'm on a call, as a part of the action items that were discussed, I might need to send out a presentation, I might need to send out a plan, I might need to send out some sort of a document to them, to the prospect to look over. Now, this is something that all thousand users of the CRM have to do once they speak to the prospect.

An AI agent, what it can do is essentially as soon as the meeting is over, an AI agent can analyze the meeting notes, identify the action items and make the whatever thing that is required to be sent, whatever follow-ups that are required to be created. It can create those tasks and it can create those documents from the information that the CRM already has from how my organization has been dealing with prospects previously.

So in my workspace of yours as CRM, I have already been engaging with let's say 5,000 prospects or 2,000 prospects. I've created these documents probably 500 times or 200 times. So not just me, other people in my organization may also have created those documents. So that AI agent can look at the action item, compare it with previous action items and the responses that other salespeople in my organization have responded to it.

and then create a document that is largely meeting the expectations of the prospect, I can very quickly edit it and send it out. So that is essentially something that an AI agent will be able to do. It will be able to also set follow-ups. So if in the call I have said that, more often than not in a sales call, a salesperson will end up saying, let's get on a call on Monday, do you have this time?

Are you available? Let's get on call next week and stuff like that. If that I created, if that is a conversation that I've had, then an AI agent can pick that up and send out a calendar invite by itself. So those are some of the things that an AI agent, this is for a CRM use case of a SaaS company. So identify, if I'm running a SaaS company, I would go in and identify all the things that are currently being manually done.

and divide that into two parts. First is where they are algorithmic and create workflows and automations within my product so that they are able to automate that. And the second is wherever there are guidelines that are required, I will build AI agents so that these salespeople are able to do it more efficiently. Does that answer your question, Kaushik?

Yeah, I was trying to find that link only. basically, because there is so much we read about having this vertical AI agent within your system and everything. But the basic understanding of how can I have it is sort of missing is what I feel. That's where the that's kind of connects to the question that Jay also mentioned, Like what is the difference between an AI application and an AI agent in the first place? So clarity itself was sort of something that was I felt like was missing in the industry.

widespread but yeah that adds up to the question. I guess with compared to simple automation even for getting certain tasks then we can set up certain automation but AI agents in themselves I mean in itself what it would do is I believe after certain iterations it would try to improve as it says it's building a bit so would that be the main advantage on you know investing on AI agents especially from you know utility standpoint.

There are a couple of advantages, not just that, there are a couple of advantages. the first advantage is that an AI agent works 24x7. It doesn't need healthcare, doesn't need leaves, it doesn't need holidays, it doesn't need any of those things. It continues to work as long as you're paying the server bills. So that's the first advantage. It never gets sick.

The second thing is that just to extend what you said, it gets trained on all the data that exists within the organization. So imagine an organization where there are 70 people who are doing cold calls every day and an AI agent can listen to all those 70 people and then get trained every day, it can become better.

It listens to all of those calls, how did they go? It will also understand and identify as to what it will be able to significantly be better at identifying patterns. You know, what happens on calls that are made to this type of customers or this cohort of people or people from this region or whatever the case may be. So that's the second advantage, which is where it can learn at scale. It doesn't need to

get training from the basics and things like that. The third advantage is that while it can learn at scale, the world is taking the responsibility of making AI agents in and of itself. Like they're learning capability also better and smarter over a period of And so there will always be organizations who will make AI agents better and smarter. And so that's the third sort of advantage.

Makes more sense, Because SaaS companies spend a lot of money, they spend more money on their people and their teams than the software itself. So, AI agents significantly creates an impact where certain areas could be done using an AI agent. Yeah, yeah, for sure.

So does this mean that it's going to take the jobs of a lot of people and would that be good for economy? is your question? So that's a great question and here's how it has been and I'll share a like a quick story with you right so in 1930s or somewhere around that time Schneider Schneider Electric which is one of the larger companies in the world

came up with lifts that closed the doors automatically. And if you can believe it, there was a union in New York called Lift Man's Union. So there is actually a union of all the lift men in New York. Obviously now the better politically correct word or more considerate word is lift people.

because there could be man or a woman. This is 1930s. So there was a lift man sort of a union and they staged a protest saying that if lift doors will automatically close and automatically open then where will we find our jobs and then over a period of time they did find jobs. So inherently

If I talk about the data that exists out there in the market, we know that AI will definitely kill 200-250 million jobs over the next five years. But then at the same time, will also create more than 300-400 million jobs. it will net, it will add more jobs than it kills for sure. Every technology advancement just makes sure that humans are working on problems that matter.

So if there's anything that is redundant, there's anything that's repetitive, technology will always try to replace that. What this also means, however, is that on the spreadsheet this makes sense always, right? A technology will take away 200 million jobs and add 300 million jobs. But from an economy standpoint, it will be a challenge for sure. Because the same people are not getting those jobs, right?

So those people have been displaced by other people and so there has to be significant amount of investment that goes into re-skilling and re-training a lot of people as to how do they work with AI agents and think of AI agents as an extension to themselves. This is something that is also very reflective in the way that we are also looking at hiring and recruitment.

Jay, Kaushik and I, have a lot of other founder friends who run SaaS companies and who run technology companies and we are seeing a trend where these companies who used to recruit are not recruiting often enough. So there are people who have 15 years of work experience and they've been working on a lot of different technologies like genuinely good experience who are finding it difficult to find jobs as of now. the 200, so while net,

AI agents will add jobs but the 200 million people whose jobs are going to be taken away over the next five years are definitely not the same people who are getting, who are part of the 300 million new jobs that are going to be added as

It's like that famous quote that is going around, it's not AI that is going to take your job, it's a person who knows how to use AI to take your job. Yeah, absolutely. Absolutely. It's a person who knows how to use AI is the person who's going to take your job because see this is and it's important for us to realize that this is the time when you know before the tsunami the water is receding. You can see that the water is receding.

You know that the water is receding. You can, you know, be complacent and stay over there knowing that inevitably something is going to come for you. Or you can see it, prepare for it and then start becoming a better, like start learning how do you work with AI agents, how can you implement AI agents and so on and so forth in your own work.

And so, Yash, could you give us an example of which industries do you think will have a significant impact with respect to having them at least in the initial stages? You can even take a bigger industry and say that like if you could tell a workflow which would help out of being in industry. So this is one of those as I shared a little bit about it earlier, right? This is one of those very few technology advancements that is going to impact knowledge workers significantly more than blue collar workers, right? So up until now, every

advancement that has happened. So let's take self-driving cars as an example. It's a threat to a blue collar worker who was working as a driver. Self-driving trucks as an example. They may not be there today but 10 years later they'll be there. so self-driving cars or autonomous vehicles are going to impact blue collar workers significantly more than it impacts knowledge workers. But AI agents are one of those very few

technological advancements where it impacts knowledge workers significantly more than blue collar workers, right? Because the very thing that knowledge workers were being paid for to process what is happening with a customer, to process what is happening with a vendor, to process what is happening with a stakeholder, a shareholder or an investor or whatever the case may be, right? Or an employee. To process the thing that is happening with the stakeholder.

is why a knowledge worker is being paid for. AI agents will actually make it faster. It means that you need one person instead of three people. Because they are now able to do that better and faster with more accuracy. So to give you an example of that is, there are use cases in HR or recruitment. Instead of having a junior HR, call candidates of interest.

who applied to your organization. Why not have AI call those people and ask your fundamental questions and see whether there is fitment. That way you can make sure that you're making calls to all the hundred people who applied for the 20 open positions that you may have. And you know for a fact that it's very well laid out in spreadsheet. doesn't get tired. It can make multiple calls at once. Essentially, it can call all hundred people at once.

instead of going one after the other and stuff like that. So a knowledge worker is getting impacted over here. But the person who gets to keep the job is the person who knows how can they deploy AI agent along with themselves, right? And then they will evolve into a person who can use AI agents and do the job significantly better than a competitive person who does not.

Interesting. So could you share some emerging trends that is happening in AI agents in general like what specific industries based on at least two to three examples of industries on how this transformation is taking place and what's the pace of it? Yeah, I mean so we've got a client right and I'll not name the client because we have an MBA but we've got a client which is into debt reconciliation or debt collection industry.

which is where an AI agent is being created and deployed to offer financing options to people who have defaulted on their loans. So imagine a bank has thousands of people who defaulted on their payments and an AI agent will essentially be very empathetic and make sure that they call them, understand what the challenges are.

why they were not able to make the payment on time and then offer them different financing options or different payment plans so that their credit score and their account can get back on track. Previously you needed a human to do that and now you can just have an AI agent do this on your behalf. Again, this cannot run with algorithm. So empathy cannot run with algorithm. Ability to find creative solutions on the call cannot run with an algorithm but it can run within a guideline.

And that's when an AI agent comes into place. Second is as I shared, recruitment is a great space, which is where you can give a guideline that these are some non-negotiable. So an organization could have a non-negotiable saying that we need all people to be work from office. We need everyone to move to the city that, like all the cities that we have, we need them to move to those cities where our offices are. We need, you know, we

We have a bring your own device concept. These are the health care plans that we can afford. These are the dental plans. These are our leave policies. These are some of the guidelines that we have. And then the AI agent can call all the people who apply, make sure that they understand that these are the guidelines. then only shortly, this actually ends up not just saving.

time and money for the company but also time and money for the candidate. They don't end up getting on calls and interviewing for a role that is not fit for them. way. So that's another industry where this is going to be a huge transformation. Another industry is FP which is Financial Planning and Accounting. can, instead of creating your own BI dashboards through an analyst or through a big data analyst or whatever

you know system that you have internally you can as a CFO or as a person who works in the finance team you can code in English so you can have an AI agent to whom you can give them the instruction that this is the chart that I'm looking for this is the insight that I'm looking for can you create that and so it will create the query and it will create a chart it will figure out all the steps it will go to the right

software, it will go to the right data sets and it will come back to you saying, hey, these things don't exist. Maybe we can fix this part of your requirement with that and stuff like that. So there are n number of industries wherever you need a person to do repetitive tasks and there are guidelines that are to be established. All of those things can be accomplished with the AI agents. Here are just a few examples. So healthcare also has examples in terms of patient records.

managing and maintaining patient records and making sure for chronic patients a lot of clinics and hospitals follow up with them in terms of how their treatment is going on on specific days and specific time depending on the treatment and stuff like that. That can happen through AI agents as well. So these sort of

You have time for one last question. But yeah, please go ahead, Sali Jai. Yeah, I was just saying that there is no limit with respect to what it could do and I guess almost all the industries fall under this. It's just you have to figure out the right use case and if you already have it implemented, you just have to see how the agents can improvise that. But the scale at which it could give results is tremendous. That's what it seems from the parents. So if I have the opportunity to go a little philosophical, I'll

I'll share this with you, right? So more often than not, we end up confusing intelligence with consciousness and those two are different things. So intelligence is the ability to process the information and figure out how something works. Consciousness is the ability to ask the why and figure out a creative solution to something that doesn't exist. So create something that just doesn't exist.

AI agents are extremely intelligent. As of the recording of this episode, they are not conscious. We don't know what happens in the future, but they are not conscious. I'd say that AI agents are essentially pushing humans to solve bigger and better problems rather than taking away the jobs or anything like that. mean, so that's essentially what it is. So they are intelligent, but not conscious.

Interesting. That sums it up I guess. Yeah, because an AI agent can show empathy but it cannot be empathetic. Yeah. It can show empathy. If you talk to, you know, model that Chad GPT has, if you talk to it, it will show empathy. Yeah. It will be able to be funny but it is not funny. And it will be a challenge. It actually becomes conscious. Then it's a different conversation, right?

then we need someone else over here talking about that what happens if AI becomes conscious. Yeah. Awesome. I think that's pretty much it from our side today. I hope this was valuable for you. We had a question, we had a conversation around AI agents for SaaS products and how you as a SaaS founder can

can implement AI agents as a part of your product and how you should think about what are the parts of your product where AI agents will be most useful for your customers as well. Thank you for joining in on this stream. And if you're watching or listening to us on either LinkedIn or YouTube, please consider subscribing or commenting or following whatever you are able to see or whatever is there on that channel.

But it was a, so hope this was valuable for you. Yeah. We'll see you until next time. Bye.