How Will AI Agents Impact Workers? | Salesforce CIO Corner

Two, three years down the lane. When we look back some of the ways how we will all be doing work, we'll be actually shocked by ourselves. How did any work get done?

Welcome to the CIO corner. Today I am joined by Prakash Kota, SVP and CIO at Autodesk. Autodesk is using AI across the enterprise, equipping its employees with all number of tools to focus on the greatest, highest value work and empowering its customers that are designing and making the world around us, from architecture to entertainment. Prakash, it's great to see you. Great to have you here, my friend.

Thank you, Juan. Thanks for having me.

We're going to be talking about how AI and AI agents are impacting the employee experience. Really an exciting part of what we do. Tell us a little bit about what Autodesk is all about, what challenges the company has faced during your transition to using more AI, certainly in the customer service side, but also in other parts of the business, and how specifically you and your team are addressing those challenges.

Across construction, manufacturing, media and entertainment, and engineering all of these areas. We bring software so that all of these creators can leverage that to design and make anything and they can also make anything in a sustainable way. What do I mean by that? With the power of AI, we feel like our creators can unlock the power of data and automate the processes that exist and avoid wastage and build solutions in a very sustainable way that can be brought to life in a physical world. And coming to your question on our challenges, it all comes down to people having natural worries on how are we leveraging data? What are the trust principles? What are the ethical principles that will be used? Is there going to be hallucinations? Should I be worried about who has access to what data? All of these things naturally are the worries that employees had when we started rolling out some of these AI capabilities. And to address that one, we have been very transparent about what we are doing and communicating that very clearly to all the different users. And also from day one pulled in a center of excellence team which included legal, security, privacy as well as appropriate business teams before we prioritized any capabilities that we brought to life. That gave them a lot of comfort. I would love for you to share to other CIOs who are just starting this journey what to keep in mind when they're embarking on an agentic AI journey.

You know, I have the advantage of being customer zero here at Salesforce, which means that we have the responsibility of taking some of these new technologies we're building onto our platform. AI technologies Agents now in particular around agent Force. And we're implementing those agents inside the company for multiple applications, whether it's customer support, customer success, whether it's. We just released a new one this week that helps our employees get better, more effective answers to technology related questions specific to the technology we support inside the company without having to send tickets to our technology support group. And we're starting to see how our employees are taking advantage of these capabilities much more effectively. It's a journey, it's not going to happen overnight. But one of the things that we've definitely learned to do now better than before is one how to go through good change management, getting people to understand how to use the technology best, how to see these agents be an extension of how they do work so that then to your point earlier, can become more efficient, more productive. And we've also gotten better at communicating effectively what the expectations are when it comes to the use of these technologies across the companies. Although those are all things that I think we all are going to have to continue to do now. One of the very important things that I think we're going to have to really get our arms around as we continue to deploy agents, as we continue to deploy AI in general is measurement, right. And the types of metrics that we would be using to be able to know how we are progressing. How are we adopting these technologies? I'm curious, Prakash, how are you within Autodesk using metrics, using measurement to drive your AI strategy?

This was one of the critical things that got a lot of anxiety to our employees as well. As you can imagine employees, some employees assumed that this was a cost cutting exercise, which it was not at all. We started focusing on employee experience, customer experience, employee productivity, how much time can we give them back so that actually they can focus on doing other important things. So when we started capturing metrics like that first, call resolution or call deflection or agentic experience, employee experience, productivity and value, when we started measuring all this, other than cost, I think that gave a lot of comfort as well to employees wanting to lean in and adopt to some of these capabilities. So, and I'll give you a couple of examples if you look at it right, like for example in customer service area, if a particular human agent was spending simultaneously, one call or one chat at a time, our goal was with all these capabilities that we are giving them to assist, can we actually handle multiple chats at the same time effectively without compromising on csat, which means faster customer resolution and more customer CSAT and more positive impact from Customer side as well as employee satisfaction also increases because their stress levels goes down. And similarly for developers, we were also focusing on it's not about how much code are you generating using some of the tools, how much are you actually accepting and checking in all the automated code that is getting in. So that talks about not only efficiency, it also talks about effectiveness. So it's been a journey. It's actually getting better and better. Every time that I see some of these metrics, it gives me a lot of hope and excitement that people are actually leaning in and leveraging some of these things.

I think that's absolutely right. You have described for our audience here the importance of having a balanced approach towards measuring the success of AI. It's not just one metric that defines how good you are at taking advantage of these capabilities. It's not just cost reduction, it's not just service improvement. It's a combination, combination of measures, but will give you a good sense as to whether the AI, the agents that you're implementing in the enterprise are actually doing what you were expecting. That is awesome. And you know, and Prakash, one of the things that's fascinating to me is, and you described it really well at the start, right? Autodesk can help you design something as simple as glass that you use for pouring water in all the way up to some of the most complex systems in buildings, facilities, computer equipment, whatever that may be. So the complexity is really, really different depending on the type of design that you need. Data has to be one of the key things here for you guys to be able to use these technologies to support that wide variety of support cases that may be coming your way. How are you treating data? I'm really curious.

So I would say data is foundational to some of these AI journeys and so it's no different to any other company. I think we are definitely putting a lot of focus on data, data standards and data governance and who has access to what. Even for our internal use cases, as an example, we are looking at what data is needed and we also don't want people to start using these proms or capabilities to get access to what they should not have. So that's where data governance plays a huge role. So we also don't want people to have easy access to ask with a prompt, some sensitive information and get access to that. On that note, Juan, I also wanted to check with you or ask you. I know you guys recently launched the Slack integration as an example to the whole agentic workforce like AgentForce and how are users getting comfortable? Because people get sensitive about what is getting shared within Slack or whatnot. But what has been initial reactions and how has that option going?

We've been in a journey here to bring AI into Slack for some time. You know, the initial deployments that we had were really focused on making available GPT type of solutions within our Einstein platform to our users for them to be able to get responses to all kinds of questions. You know, how is my deal doing? How do I close a deal? You know, questions related to benefits, questions related to the company in general, or even summarizing data within big channels where we have lots of information specific to a project or specific to an engagement with a customer. So all that has already been done. We've seen great adoption of those types of tools. Now we're building agents within Slack. The technology support agent that I mentioned before is within Slack. And one of the benefits of using agents within Slack for us is the fact that Slack is the place where people are doing work. Therefore it's easier for them to take advantage of some of the agents that we're building into Slack today. And that to me, Prakash, has really been an important takeaway. We want to make sure that at the end of the day we bring people into technologies that they're using on a regular basis and then that within the work that they do, the place where they're doing work, they can then have access to these agents and to AI.

What I'm excited about is the world ahead of us on how different it is going to be, how efficient it could be, how productive and opportunities that could raise. And when we look back of how our careers have been last 20 years or so, how different it is going to be.

You know, when I was younger and I was managing your operational groups, one of the things that was really challenging to do every morning, really early in the morning, was to go through all of these different reports and based on all these printed reports, be in a position where you would say, well, this is an area I need to tackle. This is an area that can wait. These are measures that I'm not doing well in and I need to create a plan to improve those. I 100% agree with you. In the future, all that work where it's now connected with good data plus AI and then through the use of agents, that work is going to change significantly. Giving people the ability to not spend all that time trying to understand what's in the reports, but actually doing productive work, getting people to, to, to understand how to do their jobs better, helping them in terms of better Coaching, better engagement. So I'm excited about that. I really think it's going to make a huge, huge difference. And you know, you asked me the question earlier. I think adoption is going to be key. Making sure that we have good adoption plans and that we demonstrate the value will be important.

Yeah. What I have also seen is gamification actually helps. We have actually done some level of gamification, especially for people to lean in and start changing their habits of how they do things. And when we gamify, incentivize, I think it gives a healthy competition between employees wanting to be in that wall of fame and wanting to make sure that their names are recognized and they have these modern ways of, they're adapting to the modern ways of working. And so I have seen a lot of good traction using some of these gamification techniques to really bring some speed.

I love it. I think, you know, in me doing a number of these sessions, this is the first time that we've actually spoken about how you use some of these innovative approaches to getting people to want participate more in the use of agents.

And also setting expectations on the first drone when we launch. Any of these capabilities will not be perfect because we have all seen we keep retraining and keeping focusing on the data quality. It gets better and better. The best part with agentic workflows is the agents learn much faster, have the more learning capability than humans. And so if we keep training them and updating the content, or in some cases content can also be created using agentic help. But I think it gets better and better as we have larger data sets too. So having that patience, saying that first round, it may not be perfect, is also the right mindset to lean in when we are embarking in this journey. That's something that I've also noticed.

I have a couple more questions for you because you just triggered another memory here. One of the things that you have spoken about before, and I love the term, it's the whole notion of discovery mentality. See, one of the terms that I learned many, many years ago in my previous job was the whole notion of being constructively dissatisfied, always finding ways to do things better, to improve the way things run in a constructive manner. Well, I like your term very much so, which is this whole notion that, yeah, be constructively dissatisfied, but maintain this view that you have to constantly be discovering new things. Tell me a little bit about how you apply discovery mentality within Autodesk.

Yeah, I think, See, one of the things at least I tell my leadership team and my management team and people who work within the organization is not waiting for somebody to give you a prescription or a change order for you to go focus on delivering some technology. Rather be very curious, be open minded to see what is that process, why is this used, who is the user, Be curious to know who is consuming whatever we build and then find out is there a better way to do that process? Even though a particular way was asked to us to achieve that outcome, if there is better ways to optimize that, and especially with all the help that we have now with agentic AI, I tell them to stretch and think. If there is process modernization we need to do, then probably that's what we need to focus on. Rather than automating every hundred steps we have, maybe we need only five different steps to achieve that. So why can't we strike, think and really experiment? And if it achieves the outcome we want, that's great. If not, we at least learn something and we can disregard and keep moving forward if things don't work out.

You have to have that discovery mentality to really get to the better answers, to really get to the better outcomes. As you're working on technology solutions for the company, I've been exposed to IT transformation, business transformation. You know, the whole notion of transforming the business to generate more cost savings or better P and L performance. Now we really need to be thinking about how we use this new set of technologies, agentic AI for us, agent force, to transform the way we do business across all aspects of the company. Because quite frankly, this is the first time, and this is one of the things that keeps me really excited, is that this is the first time that we have technology that really can be effective in influencing and impacting almost every aspect of the business.

100%. Can't agree more.

Well, listen, thank you so much. You've always been a great friend, a great partner, certainly a great friend of Salesforce. I'm glad that you had an opportunity to join us today. Thank you again. And please, for all of you out there, check out the links in the description below and be sure to stay tuned for more episodes of CIO Corner.