I spent some time this week in Dallas at Automation Anywhere’s Imagine 2026 conference, my second time attending the event, and once again, I walked away impressed. Not because of flashy AI demos or over-the-top claims about artificial general intelligence changing civilization overnight. We have enough of that already. What impressed me is that Automation Anywhere feels like a company that actually understands enterprise operations at a very deep level, and that understanding gives it a very different perspective on AI than many of the companies currently crowding into the market.
That distinction matters.
A lot of companies discovered AI sometime around the arrival of ChatGPT. Automation Anywhere has been building enterprise automation systems for more than 20 years. Long before generative AI, long before agentic AI, and long before every SaaS vendor suddenly rebranded itself as an “AI platform,” Automation Anywhere was helping enterprises automate workflows, orchestrate processes, integrate systems and reduce operational friction. They lived through the business process automation era, the RPA wave, intelligent automation, and now the transition into what they call APA, or Agentic Process Automation.
One thing that stands out about the company is continuity. All four co-founders remain actively involved in senior leadership roles more than two decades later: Mihir Shukla, Neeti Mehta Shukla, Ankur Kothari and Rushabh Parmani. In an industry where founders often disappear after acquisitions, IPOs or leadership transitions, that kind of stability usually says something important about the company’s culture and long-term vision.
That vision was on full display at Imagine this year. Automation Anywhere showcased new capabilities and partnerships involving NVIDIA, OpenAI, Cisco and Okta, along with new enterprise offerings built around large-scale agentic automation. The company clearly sees the future of AI not as isolated copilots sitting on top of disconnected SaaS applications, but as orchestrated systems tied directly into enterprise workflows and operations.
Still, the most important thing I heard all week was not a product announcement. It was a conversation with Mihir Shukla about what he sees as the three pillars of the Autonomous Enterprise.
I speak with a lot of CEOs, founders and analysts about AI transformation. Almost everyone agrees enterprises need to adapt. Almost everyone says organizations must rethink how they operate. But too often the conversation becomes vague and abstract. There is usually a lot of talk about “digital transformation,” “future-proofing” and “AI journeys,” but very little specificity about what companies should actually do.
What struck me about Mihir’s framework is how concise and operational it is. He boiled the challenge down into three principles that are easy to understand, measurable and, frankly, economically disruptive if organizations truly embrace them.
The first pillar is that enterprises should aim to automate roughly 80% of what they do. Not 100%. Mihir is not arguing for removing humans entirely from business operations. In fact, he explicitly believes humans need to remain involved. Human oversight, governance and judgment still matter. But automating 80% of enterprise operations would still fundamentally transform how businesses run.
That number also reframes the discussion around automation. Most organizations today think about automation incrementally. They automate a few repetitive tasks, add a copilot here and there, or improve a workflow by 10%. Useful? Absolutely. Transformative? Not really.
At 80%, automation stops being optimization and starts becoming an entirely new operating model. That level of orchestration changes staffing models, operational workflows, cost structures and even organizational design. Traditional automation technologies struggled to reach that threshold because they lacked the reasoning and contextual capabilities modern AI systems now provide. Agentic systems change that equation dramatically because they can operate across workflows, systems and decision chains in ways previous generations of automation simply could not.
The second pillar may be even more important. Mihir believes enterprises should use AI and automation to effectively 3x every employee.
I actually appreciate the realism of that target. Silicon Valley loves to talk about mythical “10x engineers,” but those conversations often drift into fantasy. A 3x productivity increase across a workforce, however, would already represent one of the largest operational shifts in modern business history.
What I found especially interesting was Mihir’s rejection of incremental productivity gains as the end goal. He specifically pushed back on the idea that 5% or 10% improvements represent true transformation. He is right about that. Those gains may improve efficiency, but they do not fundamentally change the economics or operational structure of the enterprise. They simply optimize the existing model.
A true Autonomous Enterprise requires exponential leverage. The point is not merely helping workers complete the same tasks slightly faster. The point is eliminating huge amounts of repetitive operational drag through orchestration, AI agents and intelligent automation layers so employees can operate at a fundamentally different scale. That distinction matters because many current enterprise AI deployments are struggling, precisely because they are chasing marginal improvements while absorbing massive infrastructure and software costs.
The third pillar is where Mihir’s thinking becomes genuinely disruptive. He believes the SaaS economic model itself breaks in an Autonomous Enterprise.
This was probably the sharpest insight I heard all week in Dallas.
For years, enterprises have steadily increased the percentage of IT budgets consumed by software subscriptions and seat-based licensing models. Every new operational function introduced another SaaS platform, another subscription layer and another recurring expense. Mihir argues that this model no longer makes sense in a world where automation handles the majority of workflows and AI agents increasingly execute tasks autonomously.
Why continue paying enormous per-seat licensing fees when much of the actual execution is happening machine-to-machine?
This is not simply the standard “AI will recreate SaaS applications” argument that has become popular lately. Mihir’s point goes deeper than that. He believes enterprises are dedicating far too much of their budgets simply to maintaining existing software ecosystems instead of investing in innovation, growth and research. He referenced organizations spending 75% or more of their IT budgets just “keeping the lights on.” In his view, autonomous enterprises should drive that number closer to 35% or 40%, freeing substantial capital to reinvest into the business itself.
At Imagine, several customer case studies backed this up with real-world examples. Multiple organizations discussed eight-figure and even nine-figure savings achieved through automation-led operational transformation. Those are not theoretical projections. Those are actual numbers being presented by enterprises already moving in this direction.
What also became clear throughout the conference is that Automation Anywhere does not see the current moment as a normal technology upgrade cycle. Mihir and the broader leadership team repeatedly framed this as a historic window for organizational transformation. That thinking also appears in his new book, The Five-Year Century, co-authored with Nancy Hauge. The book’s central premise is that the next five years may compress a century’s worth of business and technological transformation into an extraordinarily accelerated timeframe.
That sounds dramatic until you step back and look at what is happening around us. AI infrastructure spending is exploding. Enterprise software economics are being questioned in real time. Agentic systems are moving from experimental pilots into production environments. Organizations are beginning to rethink operational structures that have existed for decades.
Maybe “The Five-Year Century” is not an exaggeration after all.
I would also be remiss if I did not mention another aspect of Automation Anywhere that I found admirable. Neeti Mehta Shukla and her work around responsible AI and human-centered automation provide an important counterbalance to many of the fears surrounding AI adoption today. Her keynote and our interview focused heavily on how AI and automation should work for humans instead of reducing humans to servants of machine-driven systems.
Importantly, this did not feel like corporate ESG theater or carefully scripted PR messaging. It felt grounded in operational reality. As enterprises push toward increasingly autonomous systems, questions around ethics, governance and human value are not side conversations anymore. They are central business questions.
That may ultimately be one of Automation Anywhere’s biggest strengths. The company seems to understand that AI transformation is not purely a technology story. It is also an operational story, an economic story and, perhaps most importantly, a human story.
Shimmy’s take is this: Most enterprises are still experimenting around the edges of AI. They are deploying copilots, testing chat interfaces and trying to squeeze incremental efficiencies out of existing workflows. Automation Anywhere is aiming at something much larger. They are trying to redefine how enterprises fundamentally operate in an AI-native world.
Whether Mihir Shukla ultimately proves completely correct about every aspect of the Autonomous Enterprise almost misses the point. What matters is that he is articulating one of the first truly coherent operational frameworks for how organizations may need to function as AI and automation mature.
Automate 80% of operations. Use AI to 3x employee productivity. Stop devoting the majority of enterprise IT budgets to outdated software consumption models.
Simple ideas on the surface. Potentially massive implications underneath them.
After spending this week at Imagine, I suspect a lot more enterprise leaders are going to start paying attention.
