This article is part of our Future of AI series from Imagination in Action 2025 Silicon Valley Summit — where founders, leaders, and investors explored what’s next for AI. Explore the magazine.
The rise of applied AI is fundamentally changing how companies approach innovation and competition within vertical industries. It’s reshaping business operations and workflows, with companies layering AI capabilities on top of existing systems.
“This is the first time that we really have the technology that can automate knowledge work by taking in data and insights and giving recommendations,” said Bratin Saha, chief product and technology officer at DigitalOcean.
Where AI delivers immediate ROI
Industries enjoying the most success with current-generation AI models share a common characteristic: They have iterative workflows where built-in error resiliency allows for rapid improvement.
Think about drafting legal filings, processing customer service requests, analyzing financial data, ensuring sales tax compliance, creating content, or developing software. These workflows can absorb small imperfections because humans are still in the loop, but they benefit enormously from AI’s speed and pattern recognition.
“This is the first time that we have technology that can automate knowledge work by taking in data and insights and giving recommendations.”
Bratin Saha, DigitalOcean
Fields like healthcare, finance, or legal have high labor costs, “so if you can automate some of it, the ROI is there for the taking,” said Saha.
Others pointed to Klarna’s bold pre-IPO decision to sever its relationships with leading enterprise software vendors and replace hundreds of SaaS apps with an AI-powered knowledge graph. It’s an extreme example, but it illustrates the magnitude of change possible.
According to Mayfield partner Sri Pangular, the current sentiment is that “industry-specific vertical AI is the new horizontal.”
Applied AI represents unprecedented potential change for vertical software, transforming it from what might be called “sleepy SaaS” to one of the most exciting categories for building companies.
Anirudh Devgan, CEO of Cadence, agreed. Devgan emphasized that AI’s biggest impact will come from domain-specific applications, particularly because the “horizontal parts of AI will get commoditized.”
The augmentation advantage
Cost savings are one thing, but the bigger benefit lies in how AI augmentation enables entirely new business models.
Mayfield partner Sri Pangulur points to three key benefits: boosting revenue per customer, growing gross margins, and expanding total addressable markets (TAM). These, in turn, are spurring innovation in how companies price and package their offerings.
DigitalOcean’s experience with its WordPress-based Web hosting platform Cloudways provides a concrete example. The platform typically fielded roughly 600,000 annual support calls due to outages, Saha said.
By deploying AI-powered agents that quickly spot the causes of the outages and recommend fixes, Cloudways reduced support costs and expanded its TAM into proactive site management and troubleshooting. What was once a massive operational burden and cost center has become a value driver.
“Where we’re really seeing AI being helpful is as a creative accelerant.”
Michael Wise, ex-Universal Pictures
Even in entertainment, AI has a powerful role. Michael Wise, the former CTO of Universal Pictures, sees AI as either a positive or negative force.
“In my world, we’re in a moment in time where AI could be a hero or a villain,” Wise said. “A more villainous flavor is if AI is purely a cost savings mechanism,” automating visual effects and replacing writers and actors.
But, he added, “Where we’re really seeing AI being helpful is as a creative accelerant.” That includes artists using AI to create more incredible visual effects or entirely new experiences. In other words, AI also has the power to augment the power of Hollywood’s best storytellers.
Breaking down legacy barriers
Perhaps the most significant opportunity and challenge lies in integrating AI with the siloed legacy systems that power most enterprises.
Lisa Dolan, the managing director of Link Ventures, calls this the “holy grail” in the business of medicine: building AI integrations with siloed systems like CRMs, ERPs, or electronic health care records platforms like Epic.
“With AI,” Dolan says, “the big unlock is to combine these siloed pieces of data.” When health care providers can leverage consolidated information from many sources, they can enable health care providers to deliver care to patients faster than ever before.
The AI doesn’t just speed up existing processes. Bringing LLMs to hospitals will enable entirely new workflows and a more comprehensive, personalized view of patients.
The opportunity extends far beyond healthcare. Financial services organizations are integrating data from disparate sources to improve decisionmaking and serve clients more proactively. Legal teams are connecting case management systems with research databases and document archives. Manufacturers are linking supply chain data with production systems and quality control.
AI that can “read the room”
Even as applied AI reshapes business operations and workflows, founders, researchers and scientists are pursuing an even more ambitious goal of building human qualities into AI models. The aim is to make AI more understanding, more empathetic, more capable of navigating the complex dynamics of human relationships. If successful, this could transform everything from mental healthcare to materials science.
Next Chapter CEO Erick Tseng is working on an AI system that can truly “read the room.”
His mental well-being company is developing methods to “embody AI with relational intelligence.” Based on the concept of emotional intelligence, this is an effort to detect and understand interpersonal dynamics that goes beyond analyzing words.
Consider an AI couples’ therapist. Today’s AI can process what people say and perhaps detect sentiment in their language. But Tseng’s vision is to use computer vision that could detect not just sharp language between a couple, but nonverbal cues like body language, posture, eye contact—or the lack thereof—and even the sudden flush of skin.
“All these different inputs get composited into a math model,” Tseng said, “and what the AI [could] do, which no AI can right now…is say, ‘wait, timeout, I’m going to interrupt the conversation,’” allowing the AI therapist to monitor and intervene before the session goes off the rails.
Teaching AI to feel
While one application of AI is understanding human emotions, researchers are also exploring ways to give AI physical sensations similar to human touch.
Zhenan Bao, a professor of chemical engineering at Stanford, is leading work on developing new materials and sensors that can mimic the sense of touch or measure gut health in the intestine. Her lab sits at the intersection of materials science, sensing technology, and AI.
Bao’s team is using AI to analyze data from these sensors and design the next generation of sensing materials themselves, creating a powerful feedback loop. Even better, she said, they’ve found ways to get AI to predict the next set of materials they need for their work. “Our devices are really sensing systems,” she said, that “can create a lot of data for training algorithms, and in turn the AI can be used to help us understand what these data really mean.”
The upshot? Now the lab can begin to make systems that intuit things like what kind of sensing information is generated when a person touches an object? How many sensors are needed to emulate human touch? What materials would make those sensors more effective?
By answering these questions, AI is accelerating the pace of materials science research—which in turn creates better sensors, which generate better data, which improves the AI. It’s an exponential cycle of improvement.
The transformation is already underway
The adoption of AI to transform industries is no longer theory, and companies across the globe are increasingly using the technology to change the way they operate, even as AI is also helping develop physical systems that can make people’s lives better.
Though it’s still early, what was clear is that the potential benefits are enormous. Mental health services could become more accessible and effective. Materials science could advance at unprecedented speed. Human-computer interaction could become more natural and productive.
The momentum behind these efforts suggests that we’re approaching a future where AI doesn’t just process information, but it understands context, recognizes emotion or touch, and responds with appropriate sensitivity.
Explore The Future of AI | This article is part of our Future of AI series from Imagination in Action 2025 Silicon Valley Summit — where founders, leaders, and investors explored the next revolution of AI. We explored how AI is changing scientific research, creating new startup economics, straining power grids, and challenging us to rethink everything from enterprise software to regulatory frameworks. Dive into the Future of AI magazine to see the full picture.
