Table of Contents
The mindset shift needed for agentic AI success
Agentic AI is the miracle cure to all of enterprise IT’s woes — at least it is if you listen to the endless, breathless pronouncements from the industry. As with most things, there’s a kernel of truth there. Agentic AI is poised to become one of the most transformational technologies the enterprise has ever seen (breathless enough for you?), but only if we approach it properly. However, doing so isn’t about choosing the right technology or strategic direction as much as it is about getting the right mindset about it. And that may be the most challenging part of realizing true and lasting value from agentic AI.
This article is the first in a series of blog posts and webinars that aim to help you make this mindset shift. We want to help you take a fresh look at what it will take to truly realize agentic AI’s transformational potential. Every vendor wants to dive into the technology and talk about their competitive differentiation, but if you’re an enterprise leader, the place you need to start is in how you think about this new technology and how it will change how you function.
We believe that by helping you make this mindset shift — by changing how you’re looking at the technology and its role as a transformational tool — that you’ll also get a glimpse into how we’re looking at the problems that enterprise IT leaders are facing, and how we intend on addressing them. So, don’t expect a bunch of product deep dives here. This is about a vision for the future, and a belief that if we share that vision, then we can march confidently toward it, together.
What is agentic AI?
The other day, I was having a conversation with a good friend of mine. He’s a former CIO and someone whom I deeply respect as an industry thought leader. And, as I was sharing some of what we’re working on at SymphonyAI his eyes lit up. The thing is, it wasn’t so much because of what we’re doing, specifically, as much as the entire concept. “It’s called what?” he asked. He was only vaguely aware of the term agentic AI.
It’s easy to assume that everyone is as deep into all this AI stuff as I am — or, perhaps, as you are. It’s called the curse of knowledge, something I wrote about in my first book. But the fact remains that for most people, this is all brand new. So, despite all the hype, I think it’s fair to start by establishing a shared understanding of what the heck we’re even talking about here — what is agentic AI?
Of course, the challenge is that there’s no simple, universal answer to that question. In the broadest terms, agentic AI is the combination of generative AI (in the form of large language models) with the ability to leverage tools and act with varying degree of autonomy. Essentially, it’s the transformation of an interface that can merely chat and provide information into one that can act on your behalf by reasoning, breaking down complex requests into multiple actions, and then executing those actions via various other applications, automations, and interfaces.
But as you can imagine, there are lots of variations, degrees, and complications involved in actually how you bring that simple-sounding definition to life.
We’re not going to dive deep into the technological side of things in this series. The underlying “how” is critically important — and a lot of how various vendors will differentiate themselves — but they only matter once you have a deep and intrinsic understanding of what you hope to accomplish with the technology. That’s the mindset shift, and that’s where we need to place our initial focus.
Three reasons current agentic AI-first approaches are failing
Have you seen the term AI-first yet? If not, you will. It’s how we (organizations of all sorts) try to signal to the world that we’re leveraging the state-of-the-art, that we’re not hampered by the past — that we’re untethered and executing with a fresh, new AI-first perspective in everything we do.
I get it — the signaling is important — but it’s also the wrong perspective, at least from your point-of-view. As an enterprise leader, you need to be outcome-first focused.
That means that your strategy (AI or otherwise) needs to start with what you hope to achieve, the outcomes you seek. AI or any technology is merely the means to achieving that outcome. The challenge for technology companies is that they are generally coming from the opposite perspective. They have technology, and they need to find a problem to solve.
Sometimes, that works. But most of the time, it ends up putting the cart before the horse and results in misaligned priorities. I think the current focus of most agentic AI efforts falls squarely into this camp.
While there is a ton of focus on super sexy and super impactful use cases for agentic AI technologies within enterprise IT, they are falling flat for three reasons:
It’s a mess out there: When I ran large scale digital transformation efforts a few years back, I would tell executives that you couldn’t begin a transformational effort until you were nailing the basics. Transformation demands change, and no one is willing to invest in genuine change without trust that the effort will pay off. It’s the same with agentic AI. It’s a tough sell to say that you’re going to turn everything upside down when the foundations are shaky.
High risk: Building on the first point, those flashy, high impact use cases sound great on the surface. But then organizations realize just how high the stakes are and start looking at the project much more diligently. That diligence is the right thing to do, but the net effect is that it slows efforts and necessarily narrows their scope. The challenge is that if you’re going to drive meaningful organizational change, you need highly visible, easily digestible wins, which runs counter to the effort to minimize the risk on these high-risk/high-impact efforts.
Trust=adoption: Both points lead to the final issue: trust. The transformational power for agentic AI (like all transformational technologies) comes down to adoption. No technology can have a transformational impact if people don’t use it in a meaningful way. And adoption is directly related to trust, which is what the first two issues hinder. So, when tech companies go “AI-first,” lead with the technology, and target those splashy, high-impact use cases, they are hindering the adoption that will be the key to their ultimate success.
Employee productivity: agentic AI’s (first) killer app
When we looked at all of these factors, we realized that we needed to flip the paradigm upside down. Rather than starting with the technology, we asked ourselves how we could both de-risk the prospect of deploying agentic AI in the enterprise and drive trust and, therefore, adoption. When we looked at all of that in the balance, the answer was clear: our focus needed to be on the employee and helping them be more productive.
More specifically, we realized that we needed to focus on those countless activities that sap the energy and will-to-contribute from employees: mundane, administrative tasks. In short, we realized that we needed to focus on all of the things that keep employees — your team — from doing their real work. We wanted to work on leveraging our agentic AI prowess to give them more time and energy to do the very things you’re paying them to do.
Our fundamental belief is that employees want to do the challenging work that you’ve hired them to do in the first place. They want to do a good job. What they don’t want to do — what no one really wants them to do — is to spend countless hours every week on administrative overhead that may be important and essential, but that adds little value for the organization.
As we came to this realization, we knew it was the right area to focus for three reasons:
It’s universal. Administrative non-work impacts every employee. It’s fertile ground and the opportunities for massive returns are endless.
It’s foundational to your future AI efforts. The first few times your employees interact with AI in a meaningful way need to be impactful if you want to build the trust for those splashier efforts down the road. These beginning steps need to feel personal — like they’re driving value for the employee, not just the company. That’s the foundation of trust and adoption.
It’s de-risked. While the impact of alleviating the mundane administrative overhead from your employees is shockingly high — both in terms of value creation and creating a positive employee experience — the risks are also shockingly low. The consequences of a failed effort are NOT a disrupted supply chain, lost deals, or PR disasters, as they might be with splashier efforts. It’s all (huge) upside and virtually no downside.
The efficiency trap: why leveraging agentic AI to enable productivity is so hard
“If it was easy, everyone would be doing it.” – Proverb
The challenge — and irony — of this approach is that it almost seems too simple. There are a lot of tech companies talking about various parts of these challenges, but often from their AI- or technology-first perspective (like being primarily focused on their own platform ecosystem, etc.). And with so much focus on big, splashy use cases, focusing on mundane administrative tasks can seem downright pedestrian.
But as is often the case, it’s the unsexy basics that often have the most power to drive real change.
Still, there’s a deeper reason that more tech companies aren’t trying to tackle this seemingly simple challenge for the enterprise: it’s not easy. It’s very broad-based nature, its surface-simplicity, is actually what makes it so hard to tackle. In the next part of this series, we’ll get into why it’s so hard and how to fix it. Stay tuned for that.
In the meantime, take a few moments and really contemplate what might happen if you flip things upside down and don’t ask yourself how you can apply agentic AI, but instead start with your employees and ask how you can use agentic AI to drive meaningful change in their day-to-day work experience. I suspect that perspective will change everything.
