The Next Few Years..
Over the last few months, we are clearly witnessing the real takeoff of AI with direct real-world impact. This is no longer about demos or benchmarks. These systems are entering production and changing how work actually gets done. Based on the patterns visible today, here is my intuition about how the near-term future may unfold. By near term, I mean the next three to five years.
This is not a prediction or a claim of special insight. These are broad patterns that anyone closely observing AI progress can notice. You may disagree with specific points, but the overall direction should feel familiar and reasonable.
A) Enhanced human touch will become more valuable, not less
This does not mean better human to AI interaction. It means that in any complete system such as learning, healthcare, or other people-facing products; the human-to-human layer will matter more. A student may learn, practice, and get assessed by AI, but human mentors will still play a critical role in connection, motivation & mentoring. The value will not come from spending more time, but from higher-quality human involvement. Systems that ignore this will feel incomplete.
B) Software will become largely automated, with a single stack: Agentic systems.
This is already visibly happening. Products will no longer be defined by a mix of languages, frameworks, and tooling choices. Instead, they will converge toward a homogenous stack built around agents. Progress will come from giving agents more capabilities, better tools & clearer goals. Stack complexity will collapse; system intelligence will expand.
C) Direct Neural Nets, Not OS
The entire mobile & PC will be end-to-end neural networks. No OS needed in the traditional sense. You’ll just directly interact with AI, and agents will do all your tasks. It will be a single screen, meaning all your conversations will happen in one place. Agents will handle your tasks in the background and inform you on that single screen. No more switching between apps, no more navigating through folders and settings. Just one continuous conversation with intelligence.
D) Computing platform: Mobile > Smart Glasses > Neural Interface
This progression is inevitable. To get more natural and high-bandwidth interface, mobile or laptop is pretty constrained and has reached maturity. Glass will be next, then direct brain interface will be the eventual interface. Each step removes friction between thought and action. Mobile requires you to pull out a device and type or tap. Glasses keep your hands free and overlay information on your vision. Neural interface eliminates the gap entirely. Think & it happens.
E) Agents Economy
Systems of agents will take over much of the digital economy. Agents will negotiate, transact, plan, and optimize on our behalf. Humans will increasingly interact with agents rather than traditional software. Most economic activity in software will become agent-to-agent.
F) Systems must be built for both humans and agents
Future products will interact not only with humans but also with other agents. Systems must be native to both. Designing only for human UI or only for machine APIs will no longer be sufficient. Human-plus-agent interaction will become a core design constraint.
G) Super Personalization as default
Personalization will no longer be a feature. It will be the foundation. AI systems will deeply understand the people they interact with their context, preferences, habits, and intent. Generic experiences will feel inefficient and outdated.
H) Many categories will lose interfaces entirely
People may continue scrolling social media or playing games, but other categories like learning, shopping, payments etc will gradually loose visible interfaces. These activities will move into the background, triggered by intent and handled by agents without explicit interaction.
I) Ideas > Execution
As execution becomes easier and cheaper, the real bottleneck will shift to thinking. Problem framing, idea generation, and direction-setting will become the hardest and most valuable parts. Execution will increasingly be commoditized.
J) Virtual products in kinetic phase
Virtual products & experiences will go through kinetic changes and evolution. It will be very difficult to predict their form. When you’re not limited by screen size, input methods, or implementation complexity, the design space explodes. What a “product” even means in a fully virtual, AI-mediated environment is unclear. The constraints that shaped all our current thinking are disappearing.
K) Real Deployment Creates Economic Disruption
For the first time, AI is changing real economic processes in a fundamental way. Software was only the beginning. More domains will be disrupted with increasing intensity. The period around 2026-2027 will be defined by large-scale deployment across industries. Many sectors will experience their own true ChatGPT moment, not as hype, but as real structural change.