Can agentic AI do the work of 1,000 people? Are software applications as we knew them, dead? How can existing companies adapt? I’ve got reflections on all of that below, after recent conversations with leaders in the field.
If you’d like to join me – and peers – for deeper conversations on innovation and leadership, get on this list for Fortt Knox Executive Communities, launching soon: mba.fortt.com.
Uniphore
This one’s an eyebrow-raiser: Uniphore founder and CEO Umesh Sachdev told me about an industry project that shows the potential of AI agents to accelerate productivity:
We started with this big-four consulting firm. They said, “Let’s pick one domain, one customer.” We went after oil and gas. We picked a Fortune 50 company, which was their client, and the use case was contract leakage. The oil and gas company wanted the big-four firm to look at all their contracts, all their invoices, and find areas where they were not optimally set up and they were losing money. In the pre agentic AI era, the big-four consulting firm gave a proposal of 1,400 people would be required on this project for two years. We worked with them on this one client. We fine-tuned a small language model for oil and gas contract leakage. The data was synthetic. We built AI agents. Today, AI agents are running through invoices and contracts of that oil and gas company – and there are four people on the project as opposed to 1,400. This went live in production in six weeks. And the impact of that was that we went from “one domain, one client” to today, over 30 domains, over 30 small language models, and close to 100 AI agents. So it caught wildfire.
I think to clarify, the question is, on this example of four people and agentic AI replacing a proposal for 1400 people: Does it represent a viable model for the impact that AI and agentic AI can have on productivity? I know there are a lot of opinions out there about this. But this week we saw Nvidia top $5 trillion in market cap; we saw ServiceNow really have strong earnings driven by its AI agenda and strong guidance even without the federal government piece of the business being strong because of the government shutdown; and we saw the big tech underdog right now among the hyperscalers, Amazon, turn in a strong quarter for cloud. That puts real fuel behind AI valuations because there’s a sense that at least the very best, the very smartest players have an opportunity here to squeeze out a huge amount of productivity. Now, this doesn’t mean that across the board these valuations are justified. But it suggests that in some pockets they probably are. That’s enough to keep this excitement going. And it’s enough for companies to continue experimenting to continue buying those Nvidia chips and loading those hyperscaler data centers to find their own version of the right models and the right efficiency.
Khosla Ventures
Khosla Venture founder Vinod Khosla, the legendary venture capitalist, is sounding an alarm about a tectonic shift in software applications:
One of the major software vendors – I won’t name which one – their CEO asked me a couple of months ago is the idea of software applications dead? That’s a pretty potent question. And the answer is most likely yes in many, many applications. Because work happens across applications, not in applications. And work is grounded in all the knowledge, which here we’re just calling the knowledge graph, in the company – whether it’s in support or in product design, the customer doesn’t care. Work has to be good workflow across all those. And this meta operation across applications is what’s really going to be important. The applications will slowly be relegated to being less important, more like databases and storage systems. They may be systems of record, but some more efficient way to work across them is going to happen.
I think what he means is this, and I do agree with him. The way we get stuff done is going to completely shift. I think about it this way: Restaurants are dead in the sense that I’m ordering from DoorDash most of the time. Yeah, there are times when I’ll go, you know, we’ll have date night, we’ll go sit down. There’s a time when the whole family will go out before a movie or whatever we’ll do. Sure. But it used to be that takeout and delivery were for pizza, and now it’s for a lot more. Because I’ve got these DoorDash agents who are going out and doing the work of visiting the restaurant, getting the food and bringing it to me. And I think what Vinod is saying is that software is going to be similar in that there’s a different control center – control tower, as Bill McDermott at ServiceNow would put it – where you’re expressing what it is that you want and you’re not doing that in every individual application. What’s not clear is who is going to own that new interface that creates the value in the agentic workplace, because whoever owns that is going to have a lot of power, because that’s where the loyalty is going to be. And probably that’s where the software margin is going to be. The whole space is getting reconfigured, reshaped. I don’t think that means that there are going to be necessarily fewer players or fewer successful players, but the successful players will be different. Some will go away, some will be destroyed. Some will adapt properly. The smart companies are in the hunt right now to be among the survivors and thrivers.
BILL
René Lacerte at BILL announced new agents to help customers get paperwork filed more efficiently, and he’s also encouraging his workforce to adapt to AI:
AI is an equalizer across all professions and all all jobs, right? It’s an opportunity to get access to the world’s information at your fingertips. And so what we’re doing inside the company is making sure that everybody is connected and learning how to leverage AI for their jobs and their functions. We do have specific hiring around AI expertise to bring that in and help all of us get better at it. But I would say as an example, before we announced these agents, we talked about the fact that we had spent the last year or so building an agentic platform for our teams to build it on. And that is something that takes off so that the teams can build for not just this release today, but for the ability for us to iterate consistently into the future on the platform that we built. And so thoughtfulness matters. Engaging the teams, engaging employees. And what we’re seeing is that if you enable your teams to actually take advantage of the tools being built, the power of productivity is actually very high.
Right now, the advantage belongs to those who are taking initiative and figuring out a lot of this AI stuff on their own. We’re moving from a time when the power was in the hands of the artisans, the people who know how to do one or two really specific, valuable things well, and moving into the realm of the designers, those who see the big picture and understand which artisans they need to do various tasks. I was just having a conversation about this with some members of CNBC’s Technology Executive Council earlier this week, and I likened it to what I see happening with music. I make a little music on the side. I took piano lessons as a kid. I picked up guitar, but those are really the only two instruments I play. But with Logic Pro and with software instruments and a MIDI keyboard, I don’t have to know how to play flute. I don’t know how to know how to play drums, but I can actually, based on design and how I envision and imagine a song, I can put it together without having to hire those session musicians and without having to learn how to play those instruments. Myself, I think that’s where software is headed. So you’ll end up with a lot more producers, who aren’t themselves instrumentalists but are able to bring products, visions to life in a way that wasn’t possible before. But because we’re in that initial stage of that reality coming to life, even the leaders of these companies can’t tell their teams exactly how to do it. It’s going to take those design-minded people to come to the fore, design the new processes. And maybe they stay at their companies, maybe they even start their own. But I think that’s how the paradigms shift.
If you’d like to join me – and peers – for deeper conversations on innovation and leadership, get on this list for Fortt Knox Executive Communities, launching soon: mba.fortt.com.
The Uniphore case study crystalizes what the market's betting on with that $5T Nvidia valuation. Four people instead of 1,400 isn't just productivity gains, its a fundamental restructuring of how enterprise value gets created and captured. Khosla's right that we're moving from applications to orchestration layers, but the real question is whether incumbents can rebuild fast enough or if this creates the opening for entirely new platforms. The restaurnt metaphor works perfectly because just like with DoorDash, the interface becomes more valuable than the underlying service providers.