Agent-Native Metamorphism of Software
How should software products work in an agent native world
Welcome to Infinite Curiosity, a weekly newsletter that explores the intersection of Artificial Intelligence and Startups. Tech enthusiasts across 200 countries have been reading what I write. Subscribe to this newsletter for free to receive it in your inbox every week:
Hello friends,
Metamorphism refers to the change in structure or composition of a rock. It can happen due to heat, pressure, or some other natural agency. The world of software is undergoing a similar change. And it's due to the agent-native approach.
The current software products are being built using design principles from a different era. It worked well at the time. And it was appropriate at the time as well. But why on earth do we still need to memorize menus or remember shortcuts like Ctrl+Alt+M?
Say hello to today's workflow
Here's the cumbersome four-step process that I use to interact with all software:
Think of the work that needs to be done
Translate that thought into the software's language
Navigate through the maze of menus
Get the software to do the work
Why are steps #2 and #3 even a thing?
I want to tell the software what I want in plain English and have it do that piece of work.
The agent-native construct
AI agents can combine and automate steps #2, #3, and #4. Agents can proactively figure out what work needs to be done (step #1), but you still need to make the final call on what work needs to be done! Because according to Descartes, you exist because you think. If you stop thinking, you will stop existing.
Imagine a world where your software understands you, not the other way around. AI agents are on the brink of turning this vision into reality. These intelligent assistants will intuitively grasp your needs. It will eradicate the need for complex commands or navigation through convoluted menus.
Let's say you're working on a project. And instead of pausing to recall the specific steps for a function in Excel, you simply type your requirement as an English sentence. The AI agent processes your request, performs the necessary actions, and delivers the results instantly.
It's like having a personal assistant who not only understands your language, but also anticipates your needs. It learns from your habits and customizes itself to serve you better. This shift isn't just an incremental upgrade. It's a complete overhaul of our interaction with technology.
The Huffman Hint
In the field of information theory, there's an amazing body of work done by David Huffman. He is best known for his work on compressing data in a lossless way. His method can be stated simply as: If a symbol appears more frequently in your data, it should be represented with a lower number of bits.
Why can't we apply this principle to software products? Let's say I use a command very frequently e.g. calculating the average. If I use something more frequently, it should be easier to get to. But the software product doesn't make it easier. I still have to navigate to Insert > Formula > Average. Some software products are trying to do this, but it’s certainly not the norm.
Huffman's principle emphasizes the efficiency in symbol representation based on frequency of use. This can serve as a roadmap for agent-native design of software products.
AI agents can adapt to prioritize functions and commands based on individual user frequency. This means that if you frequently calculate averages in Excel, your AI agent will learn to offer this function more readily. Perhaps even suggesting it proactively!
AI agents can evolve to become not just tools, but intelligent collaborators that adapt to our working patterns. This reduces unnecessary complexity and enhances productivity.
The Battle of User Adoption: Who Will Lead?
This agent-native shift opens a battlefield where only the companies that can adapt will survive. The winners? Those who can integrate AI into their existing frameworks. Microsoft Excel is a prime example. It's ripe for an AI overhaul and Microsoft won’t let anyone else take the reins. Browser is another example. What product will be the Chrome of the agent-native era? It's going to be Chrome.
On the flip side, there are sectors without a dominant tool or a workflow. These are sectors where the knowledge professionals are perhaps dealing with a new type of work. They don't have fixed habits that need to unlearned. These sectors are wide open for agent-native products to make their mark.
If you’re a founder building in this direction, I’d love to hear from you.
If you are getting value from this newsletter, consider subscribing for free and sharing it with 1 friend who’s curious about AI: