S8E7 The Work AI Cannot See

WEBVTT
00:00:02.240 --> 00:00:07.599
Hello friends and welcome to season 8 of Inside Tech Comm with Zohra Mutabannna.
00:00:07.599 --> 00:00:16.559
This season features a collection of conversations that explore the ideas, challenges, and opportunities shaping technical communication today.
00:00:16.559 --> 00:00:22.399
From AI and content strategy to leadership, product thinking, and the future of our profession.
00:00:22.399 --> 00:00:26.399
Each episode offers a fresh perspective from people doing the work.
00:00:26.399 --> 00:00:27.760
Let's get started.
00:00:27.760 --> 00:00:32.560
Hello listeners.
00:00:32.560 --> 00:00:39.119
Today on the show we have Vinay Payapilli from Hyderabad, India.
00:00:39.119 --> 00:00:42.799
And Vinay was introduced to me by Nibbu Thomas.
00:00:42.799 --> 00:00:46.719
And Nibhu Thomas introduced me to Sagar, and Sagar introduced me to Vinay.
00:00:46.719 --> 00:00:48.079
So thank you to all of them.
00:00:48.079 --> 00:00:51.280
We are going to have a great conversation here with that, Vinay.
00:00:51.280 --> 00:00:52.000
Welcome to the show.
00:00:52.399 --> 00:00:52.960
Thanks, Zohra.
00:00:52.960 --> 00:00:54.320
Very excited to be here.
00:00:54.640 --> 00:00:54.880
Yes.
00:00:54.880 --> 00:01:05.120
Vinay, please tell us a little about yourself, what you do, and where you're at in your career, so that we can kickstart this awesome conversation that I have in mind for you.
00:01:05.519 --> 00:01:10.560
Yeah, so I've been in tech writing now for just over 26 years.
00:01:10.560 --> 00:01:19.760
And why I'm so exact about it is because I remember the exact date on which I landed in Hyderabad and then kind of like got into tech writing.
00:01:19.760 --> 00:01:22.239
So before that, I was into um instructional design.
00:01:22.239 --> 00:01:25.359
I used to write textbooks for a computer trading company.
00:01:25.359 --> 00:01:28.480
The same company asked me whether I'd they were also building a software.
00:01:28.480 --> 00:01:30.719
So they asked me whether I'd do the documentation for them.
00:01:30.719 --> 00:01:32.400
I'd never done documentation before.
00:01:32.400 --> 00:01:42.239
It was with a little bit of trepidation that I bought into it with little idea that for the next 26 years I would, you know, do it and love every minute of having done it.
00:01:42.239 --> 00:01:46.400
So yeah, here I am writing about six companies down the line.
00:01:46.400 --> 00:01:52.799
I'm totally enjoying the various, what would you call it, uh, evolutions that uh documentation has gone through.
00:01:52.799 --> 00:01:59.439
Because I remember initially we would send out Word documents and then it became PDFs that we would email out.
00:01:59.439 --> 00:02:04.959
Then you had RoboHelp come in, and then you had online help, then it became structured writing.
00:02:04.959 --> 00:02:09.199
And today we are talking about uh docs as code and things like that.
00:02:09.199 --> 00:02:15.039
So uh being part of many of those uh moves, and it's been a completely exciting career.
00:02:15.360 --> 00:02:18.240
It's so awesome to hear that you've enjoyed this path.
00:02:18.240 --> 00:02:22.560
And I think that's super important, and that's possibly why you're still in it.
00:02:22.560 --> 00:02:32.319
And we had a great conversation when we connected uh for a quick chat, and we talked about challenging some of the assumptions around documentation.
00:02:32.319 --> 00:02:42.159
And that is the approach that I would like to take with our conversation today, because you had some great perspectives that I wanted to dive into.
00:02:42.159 --> 00:02:51.520
One of the things that you mentioned was when you broke down the documentation process, you've given us the evolution in your background.
00:02:51.520 --> 00:02:56.560
But then as we look at the documentation process, you looked at the critical path.
00:02:56.560 --> 00:03:02.960
So we'll talk about what that critical path in your words meant and that writing itself wasn't actually on it.
00:03:02.960 --> 00:03:10.800
Vinet, please share with us what is it that in that documentation process, what was the critical path that you were thinking about?
00:03:11.199 --> 00:03:18.960
Yeah, so one of my subjects in college was operations research, and we had a subject where we kind of like what are critical paths.
00:03:18.960 --> 00:03:25.919
Now, for those who don't know what a critical path is, when you have any activity, you can break it down into smaller activities.
00:03:25.919 --> 00:03:34.879
In documentation, it would be, you know, like data collection, reading background stuff, testing the software, writing, review, publishing.
00:03:34.879 --> 00:03:37.120
Each of these is an individual task.
00:03:37.120 --> 00:03:41.599
And when you have all these tasks, you know what the start point and what the end point is.
00:03:41.599 --> 00:03:44.479
And many of these tasks can run in parallel.
00:03:44.479 --> 00:03:56.800
And once we put that chart, right, what task A is, what task B is, how much time it takes, blah, and you come to a path which is the longest path from beginning to end, right?
00:03:56.800 --> 00:04:10.560
And every item that is on that path, when there are two items, let's say that are parallel, one taking three days and one taking five days, and then the item that takes five days is the one on the critical path, because any delay in that activity means a delay in the project.
00:04:10.560 --> 00:04:18.639
While the activity that required three days, you you have an elasticity of about two days, and that delay does not really affect the project.
00:04:18.639 --> 00:04:21.279
So what I did was I put the critical path.
00:04:21.279 --> 00:04:22.879
So just to step back, right?
00:04:22.879 --> 00:04:32.560
Everybody, the knee-jerk reaction everybody had when ChatGPT first came out was hey, the writers are going to be out of jobs because chat GPT is going to do all the writing, right?
00:04:32.560 --> 00:04:41.600
And uh very quickly we found that ChatGPT wasn't very reliable when it came to the writing in terms of the correctness or the information it had.
00:04:41.600 --> 00:04:43.519
It was hallucinating and things like that.
00:04:43.519 --> 00:04:45.279
And everybody was trying to solve that problem.
00:04:45.279 --> 00:04:48.800
But I kind of like asked myself, hey, what's the biggest problem I have?
00:04:48.800 --> 00:04:49.680
Is it writing?
00:04:49.680 --> 00:04:52.160
That's when I sat down and I put the critical path.
00:04:52.160 --> 00:05:01.920
And I realized that the two pieces that were on the critical path, which actually delayed projects, were the data collection, getting the background information, okay, and the reviews.
00:05:01.920 --> 00:05:04.879
These were the two pieces that were on the critical path.
00:05:04.879 --> 00:05:08.959
And any delay in either of those meant a delay in the delivery day.
00:05:08.959 --> 00:05:17.199
That is when me and my team started looking to see how Gen AI could solve that problem for us.
00:05:17.199 --> 00:05:20.160
And that's what AI is pretty much very good at.
00:05:20.160 --> 00:05:23.759
Give it a lot of data sources and say, hey, give me a summary of this.
00:05:23.759 --> 00:05:25.920
And Gen AI does that very well.
00:05:25.920 --> 00:05:27.040
That was what we did.
00:05:27.040 --> 00:05:28.560
We gave all the data sources.
00:05:28.560 --> 00:05:34.480
Now, in a in a perfect world, you would have a single source of truth, and that would be something like Jira, right?
00:05:34.480 --> 00:05:36.160
Or a related software.
00:05:36.160 --> 00:05:38.160
But in the real world, that's not what happens.
00:05:38.160 --> 00:05:48.319
So some of the information is in Jira, some of it is in Confluence, some of it is in Slack or Teams, and some of it is in the Google Meet transcripts or the Zoom transcripts.
00:05:48.319 --> 00:05:55.360
So we gave all these data sources and said, hey, why don't you now go in there and give me a summary of what this feature does?
00:05:55.360 --> 00:05:57.279
What are the decisions that have been taken?
00:05:57.279 --> 00:06:00.319
Who are the people who hold key responsibilities in this project?
00:06:00.319 --> 00:06:09.519
And then every time we would rerun this skill, we would also ask it to go identify what changed from the last time it ran to the current run.
00:06:09.519 --> 00:06:11.519
And that would then tell us the what's new.
00:06:11.519 --> 00:06:22.480
Like so, if let's say the engineer has changed, or we decided to change a button to a uh, let's say a radio button, all those changes that get documented in these conversations get called out.
00:06:22.480 --> 00:06:27.920
And that was perfect because before that, we were dependent on somebody coming and telling us this.
00:06:27.920 --> 00:06:36.240
It is common for one technical writer to handle multiple features in a product, and in some companies, even multiple products.
00:06:36.240 --> 00:06:43.199
So if we were to go and attend every stand-up and read every Slack conversation, that would we would have only time for that.
00:06:43.199 --> 00:06:45.360
That would kind of like just freeze us.
00:06:45.360 --> 00:07:01.040
So having something like this that went in and basically participated in those conversations in terms of listen to everything and then came back and gave us a synopsis of what happened and the key points meant that the writers were now free to actually focus on the writing part.
00:07:01.040 --> 00:07:07.199
And more importantly, we were not dependent on somebody coming and telling us that a decision had been taken.
00:07:07.199 --> 00:07:09.600
We would get that from the AI.
00:07:09.600 --> 00:07:14.079
And then if we wanted clarity, we knew exactly whom to go and ask that question to.
00:07:14.079 --> 00:07:16.160
Hey, what does this change mean?
00:07:16.160 --> 00:07:24.319
So it took away a lot of the time-intensive stuff and let us focus on things that were key to the activity of writing.
00:07:24.319 --> 00:07:28.879
So, yeah, that's how the critical path thing kind of like rolled into this.
00:07:29.439 --> 00:07:35.920
You kind of gave us a I didn't know the textbook definition of critical path.
00:07:35.920 --> 00:07:39.360
So having that was a good thing to know.
00:07:39.360 --> 00:07:40.399
Uh so thanks for that.
00:07:40.399 --> 00:07:53.839
And then also I've also encountered in my recent day-to-day, probably in the last few months, with how that process has changed, where the upfront getting the information and then the reviews itself.
00:07:53.839 --> 00:07:58.720
In fact, I'm noticing that those bottlenecks have gone away in a sense.
00:07:58.720 --> 00:08:02.319
Which I think is empowering the profession of technical writing.
00:08:02.319 --> 00:08:06.319
So with AMI now, we know it can write.
00:08:06.319 --> 00:08:08.160
There's a lot that goes on with writing.
00:08:08.160 --> 00:08:12.000
It's not just what it's the whole process, as you talked about.
00:08:12.000 --> 00:08:14.879
There are multiple stakeholders, multiple moving pieces.
00:08:14.879 --> 00:08:16.800
So all that needs to come together.
00:08:16.800 --> 00:08:20.319
And AMI, you give a prompt, AI is just gonna do that.
00:08:20.319 --> 00:08:22.639
You need a better context and who brings that.
00:08:22.639 --> 00:08:24.399
It's the writer.
00:08:24.399 --> 00:08:31.680
We also talked about so much of the early AI conversation focused on the generating content, which you already touched upon.
00:08:31.680 --> 00:08:34.720
Whereas your team has been looking at other parts of the workflow.
00:08:34.720 --> 00:08:43.919
So if writing isn't the bottleneck, have we spent too much time asking whether AI can write documentation and not enough time asking what problems AI could solve?
00:08:44.240 --> 00:08:45.200
I think that's true.
00:08:45.200 --> 00:08:54.159
Even when it comes to writing, I think there are what we forget when we are having these conversations is that there are different types of articles we write.
00:08:54.159 --> 00:08:55.759
We don't write the same type of article.
00:08:55.759 --> 00:08:58.080
Or it's not just one type of article that we write.
00:08:58.080 --> 00:09:01.120
Let's say, for instance, I'm just taking an example here.
00:09:01.120 --> 00:09:08.639
Let's say I build a software which allows you to make payments, you know, something like GPA or something of that sort.
00:09:08.639 --> 00:09:11.360
And now, let's say it's a B2B kind of thing, right?
00:09:11.360 --> 00:09:18.159
So for one business to make payment to another business, right, instead of the interpersonal payments that we are familiar with.
00:09:18.159 --> 00:09:21.440
Now, let's say we also open a wallet, right?
00:09:21.440 --> 00:09:36.639
So we give the customer an option that, hey, you can have a wallet where instead of having multiple transactions in your bank statement, you make one large transfer to your wallet, and then all the smaller transactions take place from the wallet.
00:09:36.639 --> 00:09:45.200
Now, you can either do that, or what you can do is if you don't want to use the wallet and you don't want to use your main account, you can create your own account.
00:09:45.200 --> 00:09:46.480
So there are two ways to do this.
00:09:46.480 --> 00:09:52.879
You can use my wallet or you can transfer it to a specific account from where you are then making those transactions.
00:09:52.879 --> 00:09:56.159
Now, to document this feature, there's a couple of things.
00:09:56.159 --> 00:09:57.519
Now, how to make a payment?
00:09:57.519 --> 00:09:59.200
That's a simple document to write.
00:09:59.200 --> 00:10:03.600
And I think that kind of document can very easily be written by AI.
00:10:03.600 --> 00:10:06.480
Because if you look at code, what is code?
00:10:06.480 --> 00:10:10.000
Code is a series of steps that have to be performed.
00:10:10.000 --> 00:10:17.600
It takes some inputs from the user and some other inputs which are already in the system, which mostly are what we call prerequisites.
00:10:17.600 --> 00:10:22.799
Those are already in the system, and then there are some human inputs, then there's some magic that happens, and then there's an output.
00:10:22.799 --> 00:10:24.320
And that's exactly what code is.
00:10:24.320 --> 00:10:29.519
And AI can very well look at that code and say, hey, these are the steps you need to do to complete this task.
00:10:29.519 --> 00:10:39.279
What AI can't do is help me decide whether I should use the service provider's wallet or I should have my own bank account.
00:10:39.279 --> 00:10:42.159
What are the benefits of doing one or the other?
00:10:42.159 --> 00:10:43.919
Because everything is a choice.
00:10:43.919 --> 00:10:47.759
And that document, okay, AI can't write.
00:10:47.759 --> 00:10:48.240
Okay.
00:10:48.240 --> 00:10:59.679
So I think it's very important to understand what kind of documents AI is able to write versus what kind of documents we it can't write or the human can write better.
00:10:59.679 --> 00:11:08.000
If there is something that is how to complete a task kind of document, I would be very comfortable having an AI that's constantly looking at the repo.
00:11:08.000 --> 00:11:13.360
And every time there's a change, the AI makes a decision on, hey, does this affect the user's workflow or not?
00:11:13.360 --> 00:11:21.600
And then if it does affect the workflow, let me update the documentation and then create a PR and then the human checks it and then publishes it.
00:11:21.600 --> 00:11:27.200
But would I be comfortable with AI writing a document on, hey, why do you need this solution?
00:11:27.200 --> 00:11:29.600
How does it change your life for you?
00:11:29.600 --> 00:11:34.159
What process changes or efficiencies would it bring to you?
00:11:34.159 --> 00:11:42.240
And what changes do you have to make inside your organization to be able to benefit from the efficiencies that this software is going to give you?
00:11:42.240 --> 00:11:45.279
That is not documentation that I would trust AI with.
00:11:45.279 --> 00:11:48.320
It would require a lot of human input to be able to do that.
00:11:48.320 --> 00:11:54.000
So I think more important than can AI write or not, is what can AI write?
00:11:54.000 --> 00:11:55.919
What kind of documentation can AI write?
00:11:55.919 --> 00:12:04.080
And then once we identify that, then it becomes a lot more clearer as to do I need as many technical writers in the team or not?
00:12:04.080 --> 00:12:07.120
How much work can be outsourced to the AI?
00:12:07.120 --> 00:12:14.480
Those questions can only be answered once you're clear which of your articles can be written by AI versus which those which cannot be.
00:12:14.799 --> 00:12:23.039
The process is evolving, and as we recognize where AI can be of help or not, is your team evolving with that?
00:12:23.039 --> 00:12:28.960
And are you feeling that tech writers need to be bringing special expertise?
00:12:29.279 --> 00:12:39.519
You know, one thing that I always say, and this is a divination, but something that I always say is the two toughest roles to recruit for are the product manager and the tech writer.
00:12:39.519 --> 00:12:49.039
Now, the reason I say that is neither of these two carry over anything from their previous job to their next job unless they are moving within the same vertical.
00:12:49.039 --> 00:12:52.879
Let's say one day I'm writing for a telephone mobile company.
00:12:52.879 --> 00:12:57.600
The next day I'm writing for let's say something like an online payments software.
00:12:57.600 --> 00:13:02.399
There's very little expertise that I am carrying over from my previous job to my next job.
00:13:02.399 --> 00:13:07.200
If I'm somebody who goes in-depth into the product, yeah, I mean I can I can bring the writing stuff.
00:13:07.200 --> 00:13:07.679
Okay.
00:13:07.679 --> 00:13:09.679
But really speaking, how difficult is that?
00:13:09.679 --> 00:13:12.480
I mean, it's pretty much look at this and write.
00:13:12.480 --> 00:13:20.559
The real skill that you're looking for is is this person able to dive deep into the product that they are documenting?
00:13:20.559 --> 00:13:26.879
And are they able to understand the nuances that drive this particular industry?
00:13:26.879 --> 00:13:29.840
And that's very difficult to measure for, isn't it?
00:13:29.840 --> 00:13:35.360
So does it change the way or you know what kind of person is going to be successful?
00:13:35.360 --> 00:13:39.279
It's going to be somebody who is ready to dive very deep into the product.
00:13:39.279 --> 00:13:43.519
There are a couple of things we need to sort of like take into account here.
00:13:43.519 --> 00:13:54.639
A lot of our writing processes, I believe, are holdover from the days when we used to write for kind of like I would say household goods, like how do you use a washing machine or how do you use the oven.
00:13:54.639 --> 00:14:05.120
And I think we continue to do pretty much the same kind of writing today when we have software where the UI can be made as intuitive as possible.
00:14:05.120 --> 00:14:11.120
So I don't think we've sort of like graduated out of writing for that kind of hardware.
00:14:11.120 --> 00:14:14.559
So that's one change that we really need to look at.
00:14:14.559 --> 00:14:20.399
As writers, we shouldn't be like the guy who has a hammer and then everything he sees is a nail.
00:14:20.399 --> 00:14:23.759
We should be ready to push back and say, hey, you know what?
00:14:23.759 --> 00:14:32.399
I could write volumes on this subject, but unless the UI is fixed and the process is fixed, the user is still going to have trouble.
00:14:32.399 --> 00:14:32.879
Right.
00:14:32.879 --> 00:14:41.120
I remember that there was this one product that I used to document, and uh it was for an industry that had a lot of churn, right, in their employees.
00:14:41.120 --> 00:14:51.039
We had an employee module, and let's say you didn't want the employee to appear on the main screen, the employee's name to appear on the main screen so that he or she is assigned to a task.
00:14:51.039 --> 00:14:57.600
What you would do is typically go into the employee module and you'd search for that employee, and there's a delete button there.
00:14:57.600 --> 00:14:58.879
You'd delete their employee.
00:14:58.879 --> 00:14:59.200
Okay.
00:14:59.200 --> 00:15:04.480
And then when you go to the main screen, the employee doesn't appear on the main screen and can't be assigned to any of the tasks.
00:15:04.480 --> 00:15:05.840
Great, works for me.
00:15:05.840 --> 00:15:10.000
I don't, I'm not going to read documentation now because it's done what I wanted it to do.
00:15:10.000 --> 00:15:10.879
What did I want?
00:15:10.879 --> 00:15:15.279
I didn't want the employee to appear on the main screen and to be assigned tasks.
00:15:15.279 --> 00:15:16.320
That was my need.
00:15:16.320 --> 00:15:17.759
And I was able to achieve it.
00:15:17.759 --> 00:15:19.600
So I'm not going to read the documentation.
00:15:19.600 --> 00:15:27.200
But since I didn't read the documentation, what I didn't realize is when I delete the employee, I have deleted all this employee's records.
00:15:27.200 --> 00:15:32.240
So every payment I have made this employee for salaries and things like that, all of that has got deleted.
00:15:32.240 --> 00:15:42.399
And what we would find is at the end of the year, when the employee would ask for their tax receipts, you know, their tax sheets, whatever that needs to be filed, uh, okay, I wouldn't be able to find it.
00:15:42.399 --> 00:15:51.759
So then I would have to create a support ticket on my service provider and say, hey, can you go back into your backups and get me this information?
00:15:51.759 --> 00:16:00.159
So every March we would have the spike of tickets that came because people wanted tax information about employees who had left their employment.
00:16:00.159 --> 00:16:06.480
So and no matter how much we tried to tell them that this is the problem, they would keep saying, no, just document it.
00:16:06.480 --> 00:16:09.120
And I'd be like, it's documented, right?
00:16:09.120 --> 00:16:14.320
It's just that nobody is reading the documentation because there is no need to read the documentation.
00:16:14.320 --> 00:16:15.519
It's very intuitive.
00:16:15.519 --> 00:16:18.000
I see the employee, I see a delete button, I do it.
00:16:18.000 --> 00:16:23.440
So then what I did was we used an invisible button on top of the real button.
00:16:23.440 --> 00:16:28.720
So the user thought that they are clicking the uh the delete button, but they were actually clicking my invisible button.
00:16:28.720 --> 00:16:33.759
And there we would ask them if you delete this employee, you're going to lose all their records.
00:16:33.759 --> 00:16:35.519
Do you want to delete the employee?
00:16:35.519 --> 00:16:40.559
Or do you, or is the employee leaving the organization, in which case you have to put the last working day?
00:16:40.559 --> 00:16:42.480
So something to that uh effect.
00:16:42.480 --> 00:16:49.679
And what happened was we found that over 90% of the users actually clicked, I want to enter the last working day.
00:16:49.679 --> 00:16:54.320
And then we were able to take this to the development team and say, hey, we need to fix this.
00:16:54.320 --> 00:16:58.639
It will solve 90% of those tickets that come in March.
00:16:58.639 --> 00:17:00.720
And we were able to get it done.
00:17:00.720 --> 00:17:05.440
The thing is, you've got to be ready to do those kind of experiments.
00:17:05.440 --> 00:17:17.599
So, why I told you the story is I can write as much content as I want on how to do this correctly, but it's never going to get solved because the solution is in the product, not in the documentation.
00:17:17.599 --> 00:17:26.559
So, as documenters, it is very important that we ask ourselves: is this solution the right solution for my customer?
00:17:26.559 --> 00:17:31.279
If I put it in my documentation, does it solve the customer's problem?
00:17:31.279 --> 00:17:34.640
Or are we just washing our hands off it and saying, hey, you know what?
00:17:34.640 --> 00:17:36.880
It was there in the documentation, you should have read it.
00:17:36.880 --> 00:17:43.599
We're just transferring the responsibility from the product team or us to the customer.
00:17:43.599 --> 00:17:46.079
I don't think that's the way to retain customers.
00:17:46.079 --> 00:17:54.319
Something I like to say is my team today, we are at that point when somebody asks us to jump before we ask how high, we ask why.
00:17:54.319 --> 00:17:58.480
If you give us a good enough reason to jump, then we'll jump as high as you want.
00:17:58.480 --> 00:18:00.160
But the first question is why?
00:18:00.160 --> 00:18:01.279
Why are we jumping?
00:18:01.279 --> 00:18:06.880
And I think as documenters, that is something that we should keep in mind when we are documenting.
00:18:06.880 --> 00:18:08.880
So that's one kind of documenter.
00:18:08.880 --> 00:18:13.119
The and that relates to a completely different piece, right?
00:18:13.119 --> 00:18:21.440
And I think we discussed this on our previous conversation offline as to the key requirements of what makes a technical writer.
00:18:21.440 --> 00:18:30.559
Uh, you know, I remember when I started out way back in 2000, if you applied for a job in my team, the first thing I would look for is, hey, can this person write in English?
00:18:30.559 --> 00:18:32.880
And then the rest of it we can train them.
00:18:32.880 --> 00:18:36.319
But today, I don't care what language you write in.
00:18:36.319 --> 00:18:47.759
You could write in Gujarati for all I care, because my customer who's sitting in Japan is anyway asking the question through an AI bot, which is going to translate what you wrote into Japanese.
00:18:47.759 --> 00:18:49.759
There again, things have changed a lot.
00:18:49.759 --> 00:18:54.960
The one skill that tech writers needed to possess is no longer important.
00:18:54.960 --> 00:18:56.319
It's gone out the window.
00:18:56.319 --> 00:19:06.400
And it's now the person who can actually sit and understand the product, understand the customer, and then write in a very clear manner what is required in any language.
00:19:06.400 --> 00:19:07.759
I don't care what the language is.
00:19:07.759 --> 00:19:10.880
So I think these are the two big shifts I have seen.
00:19:10.880 --> 00:19:12.960
One is we should start pushing back.
00:19:12.960 --> 00:19:18.240
Or, you know, a lot of writing teams are pushing back and saying documentation is not the solution for this problem.
00:19:18.240 --> 00:19:20.799
And the second one is who is a documenter?
00:19:20.799 --> 00:19:22.319
I think that has also changed.
00:19:22.640 --> 00:19:30.640
I think you illustrated some of the challenges that we face and how the thinking itself needs to shift.
00:19:30.640 --> 00:19:43.200
I like the fact that you narrated this incident about where you had to push back with a solid example and how you approached it with the data backing the need, the why, why you need to shift.
00:19:43.200 --> 00:19:55.599
But I ask myself, When I how easy is it if, especially if you're a solo writer in a company, even with all the facts available to you, it's very hard to push back.
00:19:55.599 --> 00:19:58.559
The first line of failure if it is in the product.
00:19:58.559 --> 00:20:03.759
Now, I do want to call out this that I do look at documentation as part of the product.
00:20:03.759 --> 00:20:07.839
It is a product that is delivered with a product.
00:20:07.839 --> 00:20:15.119
The design is something that generally is what technical documentation is trying to solve.
00:20:15.119 --> 00:20:20.079
And what you said was no, product has to do that lift.
00:20:20.079 --> 00:20:22.240
Design has to do that lift.
00:20:22.240 --> 00:20:24.079
But that pushback is hard.
00:20:24.079 --> 00:20:30.799
As an individual contributor, probably I'm doing what I can to push back and to bring these conversations.
00:20:30.799 --> 00:20:44.160
And I think with AI now, probably these conversations are easier because if you're able to collect data factor and if you have access to those tools, then you're able to build your ask faster.
00:20:44.559 --> 00:20:44.960
Yeah.
00:20:44.960 --> 00:20:52.480
You know, I remember, and this kind of like very nicely ties into your episode with Sagar, because Sagar used to be my manager at one point, or Frank.
00:20:52.480 --> 00:21:03.440
There was this one point when the ask he had from me was when I we've got to go and reduce at least 5% of the support tickets that we that are coming in.
00:21:03.440 --> 00:21:04.720
I was like, okay, great.
00:21:04.720 --> 00:21:08.480
That's a challenging ask, and a challenge is always welcome.
00:21:08.480 --> 00:21:21.359
And I spent a whole year actually looking into support tickets and trying to figure out which of those support tickets would allow me to reduce 5% of support tickets, right?
00:21:21.359 --> 00:21:22.400
By documenting.
00:21:22.400 --> 00:21:25.519
Trust me, it's an impossible task to do it manually.
00:21:25.519 --> 00:21:40.799
Even now I can actually remember how tired I felt because every time I would find something that could reduce uh support tickets by writing a document, okay, that would push the needle by 0.5 of a percent.
00:21:40.799 --> 00:21:46.400
So 5% reduction in support tickets was just not thinkable in those days.
00:21:46.400 --> 00:21:51.599
But then you know, AI came along and AI is very good at categorizing things.
00:21:51.599 --> 00:22:14.559
So if your company has uh a support team and they have this data, they they definitely have this data, then feeding This data into AI and saying, hey, categorize this and tell me in the last 90 days which was the one question or the one solution, either ways, that was sent out the most number of times.
00:22:14.559 --> 00:22:17.200
And AI will figure that out for you.
00:22:17.200 --> 00:22:18.799
It's very good at doing that.
00:22:18.799 --> 00:22:30.480
Now suddenly it becomes easier to identify the big ticket items instead of just working on the items that have either the product manager or the customer screaming the loudest about.
00:22:30.480 --> 00:22:36.720
So in the old world, it used to be the crying child that got the milk, or in this case, got our attention.
00:22:36.720 --> 00:22:51.440
In this new world, I think we can look at it and say, hey, I want to solve this problem because the most number of customers have the problem, not because there's a particular customer who is screaming the loudest, right?
00:22:51.440 --> 00:22:54.400
And I'm going to affect the most number of customers.
00:22:54.400 --> 00:23:01.279
And maybe they've normalized this problem and they're still doing it or they are asking about it, but they aren't screaming about it.
00:23:01.279 --> 00:23:08.160
So I think in the AI world, it is easier to identify those pieces and then chase those pieces down.
00:23:08.480 --> 00:23:09.039
Very true.
00:23:09.039 --> 00:23:11.279
And I think that's probably my experience too.
00:23:11.279 --> 00:23:14.319
And there's definitely that shift that is happening.
00:23:14.319 --> 00:23:17.680
And if it is not, we definitely kind of need to look at that.
00:23:17.680 --> 00:23:26.960
Because now when that's available at your fingertips, that orientation needs to happen faster for that pushback, as we were talking about.
00:23:26.960 --> 00:23:29.440
This also kind of ties into the metrics, right?
00:23:29.440 --> 00:23:36.240
We are looking at in the traditional documentation process, we are looking at which topic we're at the most, where did we get the most?
00:23:36.240 --> 00:23:39.440
If you have feedback, then what are the pain points?
00:23:39.440 --> 00:23:48.480
But then if you're the way to address this is not uses one bug, one isolated bug from another, but looking at the overarching problem.
00:23:48.480 --> 00:23:49.279
Where is that?
00:23:49.279 --> 00:24:01.599
So even if you don't have access to the support team or that data is not solid, kind of looking at those overarching issues that will rise up, especially with AI, you can surface loan.
00:24:02.319 --> 00:24:02.960
That is true.
00:24:03.119 --> 00:24:03.279
Yeah.
00:24:03.759 --> 00:24:06.480
And uh but it also really depends on the product.
00:24:06.480 --> 00:24:24.240
I would kind of like caveat that a little bit because there are some products that unless you have access to a whole bunch of hardware and a whole bunch of data, you can't really look at it as this is what my customer is going through.
00:24:24.240 --> 00:24:26.559
Let me kind of like illustrate that with an example.
00:24:26.559 --> 00:24:28.880
Let's say you're creating a point of sale.
00:24:28.880 --> 00:24:45.440
The point of sale, and usually I find that point of sale applications are completely disconnected with the actual user because the person who takes that decision to buy that point of sale software is not the person who's standing at the point of sale and collecting that you know receipt.
00:24:45.440 --> 00:24:47.599
You know, I was at a supermarket the other day.
00:24:47.599 --> 00:24:50.640
I wanted to make a payment using a particular payment app.
00:24:50.640 --> 00:24:54.720
The guy who was checking me out said, you know, he didn't know what to do.
00:24:54.720 --> 00:25:00.880
So he looked to the lady on the next counter and asked her, Hey, how do I do this?
00:25:00.880 --> 00:25:01.519
Right?
00:25:01.519 --> 00:25:04.640
And her answer was very revealing to me.
00:25:04.640 --> 00:25:12.559
She said, Hit tat twice, press enter three times, down arrow once, a space bar, and enter twice.
00:25:12.559 --> 00:25:14.960
Okay, that was eye-opening.
00:25:14.960 --> 00:25:19.680
Because I was like, hey, this person is not worried about what's on the screen.
00:25:19.680 --> 00:25:24.079
They aren't worried about what error messages or success messages you have shown them.
00:25:24.079 --> 00:25:26.880
Okay, they are just completing a task.
00:25:26.880 --> 00:25:43.759
And unless, you know, whoever's designing that software, unless you're standing at that point of sale and you have seven people standing in queue to check out those items, you do not know the pressure that this person is going to go through when they are using your software.
00:25:43.759 --> 00:25:51.599
It might look very simple to you saying, hey, you know, I've got a drop-down list with all these items, and they've just got to select that item.
00:25:51.599 --> 00:25:57.680
But what you don't realize is in the supermarket, okay, there are like 1283 items.
00:25:57.680 --> 00:26:01.359
So your drop-down is now 1283 items long.
00:26:01.359 --> 00:26:03.680
And it is not easy for this person.
00:26:03.680 --> 00:26:10.319
It might look easy for you when you were doing it in your nice little lab, but it doesn't work in the real world.
00:26:10.319 --> 00:26:14.000
I think that is where the disconnect really is.
00:26:14.000 --> 00:26:16.880
And I'm talking about the person who's designing the software.
00:26:16.880 --> 00:26:24.640
So that really is the real-time scenario to stand there when you have seven customers waiting for you.
00:26:24.640 --> 00:26:29.359
So there was uh for one of the organizations I used to work for, we had a point of sale.
00:26:29.359 --> 00:26:35.200
And when one of our product managers was visiting one of these places, he was just sitting behind the point of sale and watching what was happening.
00:26:35.200 --> 00:26:41.519
And he saw that these people were just collecting the money and noting something down on a sheet of paper, but they weren't using our point of sale.
00:26:41.519 --> 00:26:45.279
And at the end of the day, he asked them, hey, why aren't you using the point of sale?
00:26:45.279 --> 00:26:50.960
So they were like, Oh no, your point of sale is we use it, okay, but we do it in a batch.
00:26:50.960 --> 00:26:55.279
At the end of the day, we enter everything before we close for the day.
00:26:55.279 --> 00:26:57.839
But it is too slow for us to be doing it in real time.
00:26:57.839 --> 00:27:03.759
So somebody has found a workaround for this problem, and it hasn't come to you in your support tickets either.
00:27:03.759 --> 00:27:05.599
Nobody's complaining about this.
00:27:05.599 --> 00:27:10.720
But it is a problem, and I can document all I want about how to collect a payment.
00:27:10.720 --> 00:27:12.000
They don't want that.
00:27:12.000 --> 00:27:13.680
So it is a design problem.
00:27:13.680 --> 00:27:16.480
I mean, I can document everything I want, but it's it's a design problem.
00:27:16.799 --> 00:27:21.440
It is a design problem, and I would go a little further and say it's also a market research problem.
00:27:21.440 --> 00:27:25.440
You haven't done the usability on the ground, right?
00:27:25.440 --> 00:27:30.640
The market research on the ground to really get the user research and collect that data.
00:27:30.640 --> 00:27:31.839
So that isolation.
00:27:31.839 --> 00:27:39.920
So what this tells me, Benet, is that the things that happen outside of the realm of AI still need to happen.
00:27:39.920 --> 00:27:40.559
Yes.
00:27:40.559 --> 00:27:41.279
Right?
00:27:41.279 --> 00:27:44.960
Otherwise, AI can do all the documentation you want.
00:27:44.960 --> 00:27:48.400
It's not going to solve the problem that's on the ground.
00:27:48.720 --> 00:27:49.279
Exactly.
00:27:49.680 --> 00:27:50.000
Yeah.
00:27:50.000 --> 00:27:56.559
So I think that traditional way of doing things are still very relevant.
00:27:56.559 --> 00:27:57.680
And yes, you did.
00:27:58.079 --> 00:27:59.359
I completely agree with you on that.
00:27:59.359 --> 00:27:59.839
Yes.
00:28:00.319 --> 00:28:00.720
Right?
00:28:00.720 --> 00:28:08.960
The scenario that I was talking about in a software world where you're where you have access to a product that has been well researched and you have all the information.
00:28:08.960 --> 00:28:22.559
I'm talking about probably a more not idealistic process, but more a process that is well defined, that a company that's mature and has things in place, then that's one way of looking at how do you fix it.
00:28:22.559 --> 00:28:25.119
That's not the only way.
00:28:25.119 --> 00:28:28.000
As you said, the human in the loop comes back.
00:28:28.000 --> 00:28:30.400
And because for me, that is very critical.
00:28:30.400 --> 00:28:35.119
The human in the loop is never going away with as you look at the overall process.
00:28:35.119 --> 00:28:40.640
Process may be changing, but it's the insights that are coming in still come from the ground, the reality.
00:28:40.640 --> 00:28:45.440
I have not been in that sort of a business, in that sort of a vertical.
00:28:45.440 --> 00:28:46.720
So I wouldn't have expected.
00:28:46.720 --> 00:28:55.519
Of course, in India, when we've done these payments or anywhere wherever we are doing this, there are things where the users on the ground are, they have these workarounds.
00:28:55.519 --> 00:28:58.799
And that's when it puzzles me to think what was documented?
00:28:58.799 --> 00:29:01.279
Do they even care about the documentation at that point?
00:29:01.279 --> 00:29:03.599
And the design becomes so critical.
00:29:03.599 --> 00:29:04.880
And everything before that.
00:29:05.359 --> 00:29:05.759
Exactly.
00:29:05.759 --> 00:29:10.960
And this reminds me of a very interesting scenario or something that I'd read some time ago.
00:29:10.960 --> 00:29:17.359
That when Edison came up with electricity, you know, everybody was very excited about electricity.
00:29:17.359 --> 00:29:18.720
And they all bought into it.
00:29:18.720 --> 00:29:20.960
All the factories bought into it, right?
00:29:20.960 --> 00:29:25.039
But we didn't see the benefits of electricity for another 15 years.
00:29:25.039 --> 00:29:32.000
And the reason we didn't see the benefits of electricity for 15 years is because the factories were still built for steam.
00:29:32.000 --> 00:29:38.319
And it took 15 years before those factories were redesigned to take the benefits of electricity.
00:29:38.319 --> 00:29:39.759
The same with computers.
00:29:39.759 --> 00:29:42.240
Computers came out way back what?
00:29:42.240 --> 00:29:50.000
Desktops, way back in 85, is when we started seeing desktops in offices and even some homes.
00:29:50.000 --> 00:30:01.680
But the benefits of IT didn't start to accrue until we hit somewhere in the mid-90s, is when we started actually seeing the benefits of using computers.
00:30:01.680 --> 00:30:04.480
So the same I would say with AI.
00:30:04.480 --> 00:30:13.359
AI is a great technology, but I don't think we are set up today to get or to gain benefits from what it does.
00:30:13.359 --> 00:30:18.400
And to try and even guess what that benefit would be would be futile.
00:30:18.400 --> 00:30:27.200
One of my favorite questions to uh people when they debate AI with me is you know, I ask them what the biggest benefit of the railways was.
00:30:27.200 --> 00:30:28.799
The answers are typical, right?
00:30:28.799 --> 00:30:36.319
I mean, it we move from place to place faster, we can move goods, we can move a lot more goods in one go and all those kinds of things.
00:30:36.319 --> 00:30:44.079
But to humankind, the biggest benefit that AI gave, I mean, that railways gave us was standard time.
00:30:44.079 --> 00:30:46.799
The railways couldn't run without a timetable.
00:30:46.799 --> 00:30:53.200
And until the railway started running on a timetable, every city and every town had its own time.
00:30:53.200 --> 00:31:01.920
And the biggest benefit that humankind got from the railways was that I know when it's 6 p.m.
00:31:01.920 --> 00:31:06.079
in Hyderabad, I know what time it is for Zora in her part of the world.
00:31:06.079 --> 00:31:12.559
Just because we have standard time, and there's a whole bunch of things that we can we can coordinate because we have standard time.
00:31:12.559 --> 00:31:22.319
Trying to guess what the benefits of AI is an exercise in futility, and I'm not even going to kind of like guess what my children's worlds are going to look like.
00:31:22.640 --> 00:31:27.279
Yeah, I think we have all jumped on this AI bandwagon a bit too soon.
00:31:27.279 --> 00:31:29.119
At least that's what I feel like.
00:31:29.119 --> 00:31:33.519
And you're right, the fruits take some time to ripen by the time.
00:31:33.519 --> 00:31:37.119
But right now, I feel like we are at an early stage.
00:31:37.119 --> 00:31:42.960
And too many companies are overcommitting or overexcited about it.
00:31:42.960 --> 00:31:48.160
Now, I'm an individual contributor, I'm not a C-level executive, so it's hard for me to say.
00:31:48.160 --> 00:31:55.759
But I feel, Vene, that there is a disconnect between what an individual contributor thinks and what the leadership thinks.
00:31:55.759 --> 00:32:00.160
And this is a conversation that I'm running into with my peers.
00:32:00.160 --> 00:32:02.160
Why is there such a disconnect?
00:32:02.160 --> 00:32:05.200
And this is not a rhetorical question, actually.
00:32:05.200 --> 00:32:11.599
With you leading a team and you having that access, can you please throw some light on this?
00:32:11.920 --> 00:32:15.359
So let's step into that CXO shoes for a minute, shall we?
00:32:15.359 --> 00:32:27.440
So she goes to all these conferences where she is being given demos by people saying, Hey, I built this software from scratch over the weekend.
00:32:27.440 --> 00:32:30.160
It was just me and I built the software.
00:32:30.160 --> 00:32:39.359
And your CEO is most probably looking at this and saying, Oh wow, this one person sitting in a garage somewhere built this over a weekend.
00:32:39.359 --> 00:32:49.920
And I have 500 engineers who I'm paying very good cash to, and they don't seem to be able to ship anything unless it gives me, you know, unless I give them six months or eight months.
00:32:49.920 --> 00:32:51.039
That's a very valid question.
00:32:51.039 --> 00:32:52.480
I mean, it's a human question.
00:32:52.480 --> 00:33:04.160
Recently, I had, I don't know if I should say good fortune to see this in action, where at a conference something was demoed and one of the VPs looked at it and said, Hey, we need this in the product.
00:33:04.160 --> 00:33:07.519
The director of engineering was sitting next to me and they were having this conversation.
00:33:07.519 --> 00:33:09.920
You know, the VP said, Hey, we need this in the product.
00:33:09.920 --> 00:33:13.039
And the director of engineering was like, Yeah, I agree, we need this in the product.
00:33:13.039 --> 00:33:15.200
So he was like, uh, so when can you get this done?
00:33:15.200 --> 00:33:19.119
The director of engineering said, uh, yeah, it'll take three months, I can get it done.
00:33:19.119 --> 00:33:21.920
So the VP was like, What do you mean three months?
00:33:21.920 --> 00:33:25.599
This person just built it like over a weekend.
00:33:25.599 --> 00:33:28.000
So why do you need three months?
00:33:28.000 --> 00:33:30.400
And uh, it'll take three months.
00:33:30.400 --> 00:33:31.839
Uh, there are a lot of pieces in this.
00:33:31.839 --> 00:33:34.000
So he was like, No, no, no, give it to me in 10 days.
00:33:34.000 --> 00:33:36.480
So he was like, Okay, fine, let's do it in 10 days.
00:33:36.480 --> 00:33:42.160
But let me also tell you that there are a bunch of teams we need to interact with, and I will need permissions.
00:33:42.160 --> 00:33:47.119
So, can you take responsibility for that to smooth the communication between teams?
00:33:47.119 --> 00:33:47.519
Right.
00:33:47.519 --> 00:33:49.920
And uh the VP was like, yeah, sure, I'll do that.
00:33:49.920 --> 00:33:53.440
And then the first clearance they needed was from the legal team.
00:33:53.440 --> 00:33:58.480
And uh director of engineering gave it to the VP and said, Hey, the legal needs to give us a sign-off on this.
00:33:58.480 --> 00:34:01.359
And legal took three weeks for the sign-off.
00:34:01.359 --> 00:34:10.480
Then they needed to interface with two different teams that needed to give them APIs into their pieces of code, and that took another two, two and a half weeks.
00:34:10.480 --> 00:34:15.920
So by the time this was tested and built at a scalable level, it was three months.
00:34:15.920 --> 00:34:21.119
Because a product that goes into the market can't be built unless it is scalable.
00:34:21.119 --> 00:34:26.480
You can't go to the market unless it is scalable, unless it is tested, unless it is certified for a whole bunch of things.
00:34:26.480 --> 00:34:30.159
Okay, but where did that VP's come from?
00:34:30.159 --> 00:34:31.679
And I can understand that.
00:34:31.679 --> 00:34:33.119
That's one aspect of it.
00:34:33.119 --> 00:34:38.400
The other aspect is there is a first mover advantage, right?
00:34:38.400 --> 00:34:48.719
I might do an experiment and I might fail, and I am good with that, but I'm thinking about that one experiment that might succeed.
00:34:48.719 --> 00:34:56.400
And if I don't do that experiment, or if I don't do any of the experiments, okay, I'm not going to find that one experiment that's going to succeed.
00:34:56.400 --> 00:35:03.360
As a CXO, I have my OKRs just as an individual contributor has their OKRs.
00:35:03.360 --> 00:35:03.599
Right?
00:35:03.599 --> 00:35:04.400
What are OK?
00:35:04.400 --> 00:35:05.599
And one of my OKRs?
00:35:05.599 --> 00:35:06.400
Objectives.
00:35:06.400 --> 00:35:07.840
So objectives and key results.
00:35:07.840 --> 00:35:10.000
So I'm being measured on my objectives.
00:35:10.000 --> 00:35:15.199
So as a CXO, I have my own objectives that I have to meet, my own goals that I have to meet.
00:35:15.199 --> 00:35:20.880
And one of those goals definitely at some level is hey, how are you going to save money for a company?
00:35:20.880 --> 00:35:23.679
How are we going to become more efficient in how we use money?
00:35:23.679 --> 00:35:33.039
If the question is, okay, shall I reduce my uh workforce so that I can use AI to do more with less?
00:35:33.039 --> 00:35:35.280
Okay, that is an experiment, right?
00:35:35.280 --> 00:35:38.400
And I'm ready to pay the price for that experiment.
00:35:38.400 --> 00:35:40.559
Or, you know, I might say, hey, guess what?
00:35:40.559 --> 00:35:47.519
I'm going to give all my employees, every one of my employees, $1,500 to use AI.
00:35:47.519 --> 00:35:47.920
Right?
00:35:47.920 --> 00:35:51.360
I'm going to give them $1,500 worth of tokens every month.
00:35:51.360 --> 00:35:51.760
Okay.
00:35:51.760 --> 00:35:58.480
Now, when you look at it, it doesn't make sense because $1,500, if I had $1,500, that's one headcount in India.
00:35:58.480 --> 00:35:58.960
Right?
00:35:58.960 --> 00:36:07.280
So if you're going to give $20,000 people $1,500 each, why don't you just go and hire 10,000 people?
00:36:07.280 --> 00:36:10.320
The $10,000 people are not going to push the envelope.
00:36:10.320 --> 00:36:17.199
The $1,500 in the hand of $20,000 people using AI to do their work more efficiently is going to push the envelope.
00:36:17.199 --> 00:36:22.079
So it requires you to kind of like step out of the world that you can see.
00:36:22.079 --> 00:36:23.840
It's not easy for a CXO.
00:36:23.840 --> 00:36:25.920
It's horribly bad for a CXO.
00:36:25.920 --> 00:36:30.320
Because left, right, and center, almost everybody is saying the same thing.
00:36:30.320 --> 00:36:34.719
There are companies that are gaining more customers but losing market value.
00:36:34.719 --> 00:36:41.039
And the only reason they're losing market value is because they don't seem to have an AI story.
00:36:41.039 --> 00:36:45.440
But their customer count is growing month after month after month.
00:36:45.440 --> 00:36:53.440
But somebody in the market somewhere has decided that they might be growing now, but they won't be growing in five years because they don't have an AI story.
00:36:53.440 --> 00:36:55.360
So what's the CXO going to do?
00:36:55.360 --> 00:37:02.159
The CX CXO is going to go and say, hey, give me an AI story because that's the only way I can prop up my market value.
00:37:02.159 --> 00:37:05.760
I saw a cartoon recently which is very relevant.
00:37:05.760 --> 00:37:11.519
This boss walks into the room and tells his employee, we need an AI story.
00:37:11.519 --> 00:37:15.199
And the employee looks at the boss and says, We're a restaurant.
00:37:15.199 --> 00:37:17.519
So he says, I don't care.
00:37:17.519 --> 00:37:18.880
We need an AI story.
00:37:18.880 --> 00:37:23.840
So the employee quickly writes something on a piece of paper and then turns it around to the boss.
00:37:23.840 --> 00:37:25.599
It says, DOSI.
00:37:25.599 --> 00:37:31.920
So instead of dosa, it says dosai, and AI is kind of like a little separate from the DOS.
00:37:31.920 --> 00:37:32.880
So it's dosai.
00:37:32.880 --> 00:37:38.960
We have an AI story now, and the boss is happy because everybody needs an AI story.
00:37:38.960 --> 00:37:49.760
Finally, I will wrap up this answer with an insight that I got when I was talking to a CXO about they just had a round of layoffs and I was talking to him.
00:37:49.760 --> 00:38:00.800
And he said, You know what, when I looked around the organization and I identified the people who don't understand what AI can do, who are not ready to experiment with AI.
00:38:00.800 --> 00:38:03.119
They aren't pushing the envelope.
00:38:03.119 --> 00:38:06.239
Now, those people I don't want in my organization.
00:38:06.239 --> 00:38:08.880
There is no place for them in my organization.
00:38:08.880 --> 00:38:10.880
And I completely get that.
00:38:10.880 --> 00:38:18.480
Because any CXO is going to be looking at their team and saying, hey, that one brilliant idea lies in that team.
00:38:18.480 --> 00:38:29.840
But it lies with the people who are working with AI, not the people who are working against AI or who are refusing to see what possibilities AI is going to open up for them.
00:38:29.840 --> 00:38:32.960
So I get where your question is coming from.
00:38:32.960 --> 00:38:38.639
And I guess the answer is nobody really knows what AI is going to do.
00:38:38.639 --> 00:38:50.159
But they are every CXO is looking for people who are playing around with AI and pushing that envelope on AI because every CXO believes the answer lies in their team.
00:38:50.159 --> 00:38:55.360
I am one of those people who could possibly have a solution that will allow you to leverage AI.
00:38:55.360 --> 00:38:56.719
I hope that answers the question.
00:38:57.039 --> 00:38:58.960
It's definitely revealing to me.
00:38:58.960 --> 00:39:01.360
I do come from a place.
00:39:01.360 --> 00:39:08.559
I understand, Vene, the challenges that the CXO is dealing with or grappling with.
00:39:08.559 --> 00:39:10.559
But there is still that disconnect.
00:39:10.559 --> 00:39:17.920
You brought up a good point that if I'm not thinking creatively about AI or if I'm against it, then I may not be relevant to an organization.
00:39:17.920 --> 00:39:19.199
I understand that.
00:39:19.199 --> 00:39:23.920
But there is, again, I'm speaking from an individual contributor perspective.
00:39:23.920 --> 00:39:32.960
There are many of us that around me that have been laid off who are good, who are amazing, who are creative, and yet they were let go.
00:39:32.960 --> 00:39:34.960
Nobody knows what AI is going to do.
00:39:34.960 --> 00:39:37.039
Nobody knows what the future is.
00:39:37.039 --> 00:39:40.559
But it kind of runs counter to what you talked about.
00:39:40.559 --> 00:39:48.719
There are peers who are not experimenting with AI, and the CEO doesn't have visibility into that.
00:39:48.719 --> 00:39:50.159
How do we communicate?
00:39:50.159 --> 00:39:53.440
I think my question here is how do we bridge that gap?
00:39:53.440 --> 00:40:03.360
Because good talent is being locked in this whole AI-first rush to experimentation and yeah, I get that question.
00:40:03.519 --> 00:40:13.679
And the fact of the matter is that it is very difficult to individually look at people and say, okay, is this person AI ready?
00:40:13.679 --> 00:40:15.119
Is this person not AI ready?
00:40:15.119 --> 00:40:21.360
So more often than not, what you will find in organizations is a complete theme or a complete vertical being laid off.
00:40:21.360 --> 00:40:36.960
There might be a bunch of people who are very good at AI, but they just get lost because they are part of a vertical that the organization at some level has decided can be rolled into yet another vertical or into yet another theme, right?
00:40:36.960 --> 00:40:38.000
It's just that.
00:40:38.000 --> 00:40:41.760
I mean, it is not a reflection on the person per se, right?
00:40:41.760 --> 00:40:42.880
It's just that.
00:40:42.880 --> 00:40:47.599
And the fact of the matter is we are very good at reinventing ourselves.
00:40:47.599 --> 00:40:51.840
I mean, you know, from most of your listeners, maybe from outside India.
00:40:51.840 --> 00:40:55.199
So let me kind of like lay out the scenario that I'm talking about.
00:40:55.199 --> 00:41:01.280
About 20, 20, 30 telephone boots, manned telephone booths all over India.
00:41:01.280 --> 00:41:04.719
And uh uh that was a very steady source of income for a lot of people.
00:41:04.719 --> 00:41:07.920
Having a telephone booth uh was a steady source of income.
00:41:07.920 --> 00:41:15.440
And then mobile phones came along, and suddenly there was a mobile phone in everybody's hand, and the telephone booth just disappeared.
00:41:15.440 --> 00:41:19.920
Now, the question really is what happened to all those people who were running the telephone booth?
00:41:19.920 --> 00:41:20.400
Right?
00:41:20.400 --> 00:41:21.920
They were they were good business people.
00:41:21.920 --> 00:41:26.159
They were running their businesses well, they were setting aside money for tomorrow or whatever, right?
00:41:26.159 --> 00:41:33.679
They all basically evolved, and today many of them are running mobile phone peripheral shops.
00:41:33.679 --> 00:41:39.199
They give you the back covers for mobile phones and the screen guards and the charging wires and things like that.
00:41:39.199 --> 00:41:49.679
I know that that's a very simplistic example of uh things, but the fact of the matter is AI at one level is taking away jobs or people are losing jobs.
00:41:49.679 --> 00:41:59.360
But according to me and to whoever I speak to, that's going to change very soon because the number of companies is going to explode, right?
00:41:59.360 --> 00:42:14.719
Because what AI allows us to do is it allows us to take an idea from where it is just a thought to a fully fleshed out working prototype, more than a working prototype, in much faster time than we could ever do.
00:42:14.719 --> 00:42:22.320
So that does not mean that we are going to now build the same amount of software that we used to build, say, five years ago.
00:42:22.320 --> 00:42:27.679
Instead, I think the number of software that's going to get built is going to explode.
00:42:27.679 --> 00:42:34.880
And as the number of software explodes, you are going to need people to build those software.
00:42:34.880 --> 00:42:39.519
And that is going to mean that there's going to be an explosion of jobs at some point or the other.
00:42:39.519 --> 00:42:46.639
It's just that we are kind of like going through that dip right now where people haven't yet decided what they want to do.
00:42:46.639 --> 00:42:49.920
So I don't think that this is a permanent situation.
00:42:49.920 --> 00:42:57.119
I believe that this is very quickly going to change where people start seeing new ideas and experimenting with new ideas.
00:42:57.119 --> 00:43:03.599
And as those new ideas come to fruition, we are going to get jobs and we're going to get a lot more jobs.
00:43:03.599 --> 00:43:06.159
Whether they're going to be exciting jobs or not, I don't know.
00:43:06.159 --> 00:43:09.920
You know, we might end up being delivery boys for food.
00:43:09.920 --> 00:43:11.039
I don't know.
00:43:11.039 --> 00:43:13.360
That might be what it allows us to do.
00:43:13.360 --> 00:43:15.599
But there are going to be a lot more jobs eventually.
00:43:15.599 --> 00:43:18.000
It's this is something that is temporary.
00:43:18.000 --> 00:43:20.159
I don't see this going for more than a couple of years.
00:43:20.480 --> 00:43:27.199
I love that optimism and the message that the human spirit endures and will overcome.
00:43:27.199 --> 00:43:37.599
The only thing that I'm seeing, you took me down memory lane with the PCO booths in India, what, 20, 30 years ago, before mobile phones became commonplace.
00:43:37.599 --> 00:43:40.880
The pace at which AI is changing is faster.
00:43:40.880 --> 00:43:44.800
And I think the disruption is more stark.
00:43:44.800 --> 00:43:51.360
And the impact on human lives, which I think is the unfortunate downside that it's happening too fast.
00:43:51.360 --> 00:43:55.760
And not too many companies are coming with great ideas too fast either.
00:43:55.760 --> 00:43:58.559
So, but I do see this disruption.
00:43:58.559 --> 00:44:07.760
And when I say disruption, Disruption, not the CXO disruption definition, the negative impact of a disruption that's happening with the layoff.
00:44:07.760 --> 00:44:09.519
But then this question comes to my mind.
00:44:09.920 --> 00:44:15.679
If I could just stop you, if I could just kind of like interject over here, sorry, but you know, let's go back to Sadar, right?
00:44:15.679 --> 00:44:19.840
And Xenius is a product that he built with his team in two years.
00:44:19.840 --> 00:44:23.280
A product like Xenius could never have been built in two years.
00:44:23.280 --> 00:44:26.400
It would have taken a lot more time than that.
00:44:26.400 --> 00:44:29.760
You know, simpler products than that have taken a decade to build.
00:44:29.760 --> 00:44:36.159
So there are going to be people like Sagar who are going to come up with ideas based on their experiences.
00:44:36.159 --> 00:44:39.519
And they are going to say, hey, it's now easy for me to build it.
00:44:39.519 --> 00:44:40.559
Let's take a look at it.
00:44:40.559 --> 00:44:41.679
You know, two years.
00:44:41.679 --> 00:44:52.880
If you and I, we've set aside money because we want to start some company or we want to build a product, and we know that it's going to take two years versus we know that it's going to take 10 years.
00:44:52.880 --> 00:45:04.239
Can you imagine the fewer number of entrepreneurs there will be if a product is going to take a decade to build versus if a product is going to take, let's say, a year to build?
00:45:04.239 --> 00:45:04.559
Right?
00:45:04.559 --> 00:45:09.920
I just need to set aside money to support me for a year, and then I will have a full-fledged product.
00:45:09.920 --> 00:45:18.800
So the number of entrepreneurs is going to go up just because it's easier to save up Monday for tomorrow and take the risk today.
00:45:19.119 --> 00:45:19.440
Gary.
00:45:19.440 --> 00:45:19.920
Gare.
00:45:19.920 --> 00:45:20.880
So yeah.
00:45:21.440 --> 00:45:22.559
It's very optimistic.
00:45:22.880 --> 00:45:26.079
Yes, I would like to go with that optimism as well.
00:45:26.079 --> 00:45:42.800
And like you said, I think I was reading somewhere that the gig economy economy is just going to like skyrocket with the avenue that you and people are hustling and people are able to, those who've been laid off, I what I read on LinkedIn, people are thinking creatively.
00:45:42.800 --> 00:45:43.679
How do I survive?
00:45:43.679 --> 00:45:45.519
How do I write this out?
00:45:45.519 --> 00:45:48.320
And they are definitely thinking creatively out of the box.
00:45:48.320 --> 00:45:50.400
So yeah, there is something to be said about that.
00:45:50.400 --> 00:45:58.000
I guess what I'm trying to do is invite my listeners to think along those lines that this will not last, the bad times will not last.
00:45:58.000 --> 00:46:02.960
And to think about if they've had an idea to leverage AI, if they've been against AI.
00:46:02.960 --> 00:46:08.239
So I think I'm kind of trying to challenge that notion that maybe, maybe it's not another job.
00:46:08.239 --> 00:46:11.360
Maybe it's something a creative idea that you've always lived with.
00:46:11.360 --> 00:46:13.360
Maybe it's time to experiment with that.
00:46:13.360 --> 00:46:15.039
I love your optimism, Vene.
00:46:15.039 --> 00:46:17.199
I love how this conversation.
00:46:17.199 --> 00:46:20.320
I appreciate you taking these hard questions.
00:46:20.320 --> 00:46:25.519
I have not attempted this before, but I don't get to talk to senior leadership.
00:46:25.519 --> 00:46:28.400
So you kind of were my segue into that.
00:46:28.400 --> 00:46:30.800
And you gave me a different perspective.
00:46:30.800 --> 00:46:35.840
You know, that human touch again on how I should probably shift my thinking.
00:46:35.840 --> 00:46:40.159
It hasn't fully shifted yet, but I think that dialogue needs to happen.
00:46:40.159 --> 00:46:43.519
And I appreciate that the insights that you've brought to it.
00:46:43.519 --> 00:46:48.559
Any last thoughts that you would like to touch upon or add to the conversation that we've had?
00:46:48.960 --> 00:46:53.440
I would just like to finish off with a lot of people ask me when I what should I do to be ready for tomorrow, right?
00:46:53.440 --> 00:46:55.360
Or for today, uh, for that matter of fact.
00:46:55.360 --> 00:46:58.639
And I'm like, there is nothing like being ready.
00:46:58.639 --> 00:47:00.800
You can be ready for what you can expect.
00:47:00.800 --> 00:47:06.880
We are in you know, in space that we have never really encountered before.
00:47:06.880 --> 00:47:12.159
So all you can do is to be open to the ideas that are flowing towards you.
00:47:12.159 --> 00:47:19.119
Don't be close to those ideas and dive deep because uh expertise is going to be valued.
00:47:19.119 --> 00:47:25.360
Today, AI can take a picture and uh redo it for you the way Van Gogh would have done it.
00:47:25.360 --> 00:47:31.679
But then uh, do you value that painting as much as uh if Van Gogh had really done it?
00:47:31.679 --> 00:47:37.199
So people who can actually create that is going to really have value.
00:47:37.199 --> 00:47:45.440
So diving deep and not looking at your job as just a way to earn a salary, but uh, you may not get the perfect job.
00:47:45.440 --> 00:47:52.480
But once you're in there, I think it is very necessary that you dive deep and understand the space in which you are.
00:47:52.480 --> 00:47:56.480
If you have that attitude, I think you'll survive anywhere.
00:47:56.480 --> 00:48:04.320
So that is the advice I give anybody who asks me what they should do to prepare for the world that uh we find ourselves on.
00:48:04.559 --> 00:48:05.519
Thank you, Benet.
00:48:05.519 --> 00:48:07.280
This was a fantastic conversation.
00:48:07.280 --> 00:48:08.800
I've learned so much.
00:48:08.800 --> 00:48:12.159
I appreciate everything that we've talked about.
00:48:12.159 --> 00:48:14.000
Thank you again and have a wonderful day.
00:48:14.000 --> 00:48:16.320
Thank you.
00:48:16.320 --> 00:48:19.440
Thank you for listening to Insight Techcom.
00:48:19.440 --> 00:48:23.760
If there's a guest you would love to hear on the show, please let me know.
00:48:23.760 --> 00:48:29.039
And don't forget to follow and catch every episode on your favorite podcast app.
00:48:29.039 --> 00:48:30.400
See you next time.