📺 Today’s recommended deep-dive video: https://www.youtube.com/watch?v=gbDl28Hx9TA
The $10 Million Domain and the Future of ChatUX
Generative AI is not just a tool; it is the single largest tech paradigm shift since the internet, surpassing the impact of the mobile revolution. HubSpot co-founder Dharmesh Shah argues that we are moving toward a “declarative” world where knowing what you want is more important than knowing how to code.
Core Question: How will the shift from graphical user interfaces to natural language “ChatUX” redefine software and entrepreneurial opportunity?
Highlights
- The transition from imperative (step-by-step) to declarative (result-based) software models.
- Why “vector embeddings” represent a billion-dollar opportunity for semantic matching.
- The “App Store moment” for AI through the launch of ChatGPT plugins.
- Dharmesh’s $10M+ purchase of chat.com as a bet on the future of human-computer interaction.
⏱️ Reading time: approx. 12 minutes · Saves you about 65 minutes vs. watching.
Want to take notes while watching? Click the image below and let AI Notebook capture the key points for you 👇
The Declarative Shift: Software That Understands Intent
From Clicks to Conversations
For thirty years, we have lived in an “imperative” software world. If you wanted a report in a CRM like HubSpot, you had to navigate a series of clicks, drags, and swipes, executing a manual sequence to achieve a result. You had to learn the software’s language to get what you wanted.
Generative AI flips this model on its head.
We are entering the era of the “declarative” model, where the only requirement is knowing what you want and being able to express it in plain language. If you can speak English—or any language—you can now command complex systems to perform tasks that previously required specialized certification or coding knowledge.
This isn’t just a fancy version of autocomplete.
While critics argue that Large Language Models (LLMs) are simply predicting the next word, Dharmesh argues they are actually “reasoning engines.” They take a snapshot of human knowledge and apply logic to solve problems in real-time. When a user tells an AI “my friend can’t see my website,” the AI doesn’t just look for those words; it reasons that the user is likely missing a hosting provider or a specific folder structure, acting as a senior partner rather than a junior intern.
💡 Digging Deeper
Q: Is the goal to eliminate the need for coding entirely?
A: Not necessarily, but it lowers the floor. Natural language becomes the interface, allowing non-technical users to generate code, troubleshoot errors, and deploy applications like websites in minutes.
Q: Why does Dharmesh compare this to the birth of the internet?
A: Because unlike mobile, which had discrete use cases (GPS, cameras), AI impacts every industry simultaneously by changing how we interface with all information.
Q: What is “ChatUX”?
A: It is a chat-based user experience where the software acts as an all-knowing assistant that handles the “how” so the human can focus entirely on the “what.”
The Math of Meaning: Vector Embeddings
Mapping Human Thought to Geometry
To understand why AI is so powerful, you have to understand vector embeddings. Imagine every paragraph, tweet, or blog post reduced to a single point in a space with a thousand different dimensions. While humans can only visualize three dimensions, AI uses math to plot the “meaning” of text as a coordinate in high-dimensional space.
This allows us to calculate “semantic distance.”
In a traditional search, you look for keywords. In a vector-based system, you look for proximity of intent. If one person writes about “founder struggles” and another writes about “loneliness in leadership,” a keyword search might miss the connection, but a vector embedding recognizes they are mathematically close.
This technology allows for near-perfect matching in industries like dating, recruiting, or community building. By converting raw text into these mathematical vectors, businesses can find hidden patterns of compatibility that humans—and traditional databases—would never see.
💡 Digging Deeper
Q: What is a Vector Database?
A: It is a specialized storage system, like Pinecone, that allows companies to store these mathematical representations of text and search through them at lightning speed.
Q: Can this work if people are lying or being inauthentic?
A: With a large enough sample size, AI pattern matching often uncovers inauthenticity because the mathematical “fingerprint” of the text doesn’t align with known successful or honest patterns.
Q: Is this hard to build?
A: Dharmesh notes that “near mortals” can now build vector embedding models over a weekend using tools like Python and OpenAI’s APIs.
The $10 Million Bet on chat.com
The App Store Moment for AI
The announcement of ChatGPT plugins is the “iPhone moment” for the AI era. Just as the App Store turned a phone into a platform for millions of specific utilities, plugins allow ChatGPT to interact with the live internet. It can now book your flights via Expedia, update your CRM via HubSpot, or order your groceries.
This transforms AI from a smart chatbot into a smart ecosystem.
Dharmesh recently purchased the domain chat.com for a reported eight-figure sum (over $10 million). This wasn’t a corporate HubSpot move; it was a personal bet on the fact that “Chat” will be the primary interface for the next two decades of technology. It is a “cover charge” for the biggest party in tech history.
He isn’t just buying a name; he is buying a seat at the table.
For entrepreneurs, the opportunity lies in the intersection of “niche expertise” and “AI utility.” You don’t have to build the next LLM to win. You simply need to take a specific community or problem—like founder peer groups—and apply AI to solve the “experience” rather than just the “transaction.”
Key Takeaways
The most successful entrepreneurs of the next decade will be those who embrace “Prompt Engineering”—the art of speaking to AI to get specific, high-value results. This isn’t just a technical skill; it’s a communication skill. It requires the ability to chain together different AI tools, like using Whisper for transcription, GPT for summarization, and custom prompts to extract frameworks and ideas.
Don’t be a grifter looking for a quick arbitrage. Instead, look for real problems where “keyword matching” is failing and “semantic meaning” could provide a 10x better solution. Whether it’s through vector databases or custom ChatUX, the goal is to create lasting value, not just a wrapper for someone else’s API.
Finally, have the courage of your convictions. Dharmesh waited 17 years for the technology to catch up to his vision of a natural language business assistant. When the paradigm shift finally arrived, he didn’t hesitate to clear his calendar and invest millions. In a world of exponential shifts, the greatest risk is not being curious enough to play with the tools.
Q&A
Q1: Why did Sam Altman reportedly refuse equity in OpenAI?
He reportedly felt he already had enough wealth and didn’t want the financial incentive to cloud his judgment during the development of potentially world-changing technology.
Q2: Is AI going to take everyone’s job?
Dharmesh views it as “Human to the AI power”—an amplifying force. While some roles will be eliminated, new roles will emerge that create more net value for the world, similar to the transition during the birth of personal computing.
Q3: What is LangChain?
It is a popular open-source library that helps developers “chain” together different Large Language Models and data sources, making it easier to build complex AI applications.
Q4: How did Shaan Puri use AI for the podcast intro?
He used ChatGPT to write a rap, then used a Google Colab folder and a voice model to transform his own voice into a perfect replica of Kanye West rapping the lyrics.
Q5: What is the difference between Imperative and Declarative models?
Imperative requires step-by-step instructions (“how” to do it), whereas Declarative only requires a description of the desired outcome (“what” you want).
Q6: What is a “Reading Week”?
Inspired by Bill Gates, it’s a practice of clearing one’s entire schedule to immerse oneself in a single topic—like AI or Crypto—to move past surface-level knowledge and build true intuition.
Q7: Will Dharmesh leave HubSpot to do AI full-time?
No. He has designed his role at HubSpot to have no direct reports and no administrative meetings, allowing him to tinker on experimental projects like ChatSpot while still contributing to the company.
