Didn’t Discover!!
What You're Looking For?
- Services
-
Didn’t Discover!!
What You're Looking For?
By
Aman Bhati
6 min
When Dharmesh Shah, co-founder and CTO of HubSpot, walked onto the main stage of INBOUND 25, the audience had just returned from lunch and was still buzzing over Yamini Rangan and Karen Ng’s HubSpot Spotlight session.
After a warm welcome from the audience, Dharmesh started his session on “You to the Power of AI” with a clear central message: the future of work is not humans competing against AI but humans achieving more with AI.
He emphasizes that AI is not here to compete with us but to amplify us. If you’ve been wondering how to use AI practically without losing the “human” in the process, this session was packed with answers.
Dharmesh began his session with a simple but powerful experience. A few days back, he posted a LinkedIn poll asking, “How should humans compete with AI?”
The responses were telling:
This split of answers revealed the fear of being replaced by AI vs the excitement at being enhanced.
Then, he reframed the conversation: we don’t need to fight AI. We need to build with it. AI is not an adversary; it’s a partner. Like we once learned to use calculators, computers, and the internet, AI is the next leap in augmenting human potential.
Understanding the fundamentals is the first step toward confidence with Artificial Intelligence.
Dharmesh continued the session and explained the mechanics of generative AI in plain language. At its core, a large language model (LLM) is designed to predict the next token in a sequence of text. A token is simply a small chunk of language, sometimes a word, sometimes part of a word. On average, each token represents about three-quarters of a word.
This may sound like autocomplete on steroids, and in many ways it is. But with billions of parameters tuned during training, the predictions become sophisticated enough to produce articles, stories, computer code, and analysis that can feel indistinguishable from human thought.
At the end of this part, he jokingly added that if autocomplete in your phone is like a high school student guessing the next word, LLMs are autocomplete with a PhD in nearly everything.
Many professionals use AI every day without caring how it works. But even having some foundational knowledge will keep you going a long way with AI.
He linked it to his childhood experience of learning to play a Casio keyboard. For years, he learned songs by trial and error. Only later, after studying music theory, did he realize how much faster he could progress with the proper foundation. The same applies to AI: understanding a few basics helps people unlock far more value.
Despite being powerful, AI comes with some important-to-know limitations, like:
To overcome these cons, Dharmesh emphasized the importance of providing accurate context. Every helpful instruction, supporting resource, or tool must be passed into the context window, the model’s working memory space.
PART means Prompt, Archive, Resources, Tools, and they act as the building blocks of context.
These elements determine whether AI gives a generic or highly tailored answer. As Shah said, “In the world of AI, context is queen.”
As new AI assistants are released almost monthly, many professionals feel overwhelmed about which to choose. Dharmesh simplified the choice and suggested sticking to a leading frontier model like GPT-5, Claude, or Gemini. All three are powerful and creative enough for most businesses around the world.
Not just that, the real success also depends less on your chosen model and more on how you use it.
So next, he introduced a simple formula for using AI at work:
Besides the 60-30-10 rule, he also suggests practicing meta-prompting to win with AI. Ask AI to improve the prompt you use with it. Provide accurate clarifications like target audience, tone, and preferences, and ask AI to convert them into a decent and better prompt.
Also, provide custom instructions to the AI tool, in which you brief it about how you want the tool to behave. Input everything from tone of voice to formatting to persona to boost consistency throughout responses.
A new open standard that makes it easy for AI to connect with different tools, like a universal USB-C port. HubSpot already supports this, making AI feel less like an isolated chatbot and more like a teammate plugged into your workflow.
Finally, before starting any task, ask yourself, “Can AI do this?” or “Just try it with AI.” Even if it doesn’t work today, chances are it will work tomorrow, or at least in three months when the models improve.
As he continues the session, Dharmesh reflected on his prediction at INBOUND 24: that the coming year would be the “year of AI agents.” On stage this year, he corrected himself with a smile: “It will be the decade of AI agents.”
Even HubSpot’s agent.ai platform illustrates this momentum. It grew from 47,000 early users to over 2 million in just one year. Even more encouraging, 26,000 of those users have built their own custom agents, with hundreds shared publicly across industries.
All this is just the beginning; soon, agents will become the digital teammates of future collaborators who extend human capacity rather than compete with it.
Many companies rely on AI heroes, individuals who experiment heavily and discover useful AI tools. But besides these trailblazers, Dharmesh argued that true transformation comes when those insights scale across teams. That’s where his TEAM framework comes in:
This loop ensures that instead of relying on one or two AI enthusiasts, entire teams adopt repeatable habits that scale.
Our strongest advantage is our emotional intelligence. AI can handle repetitive tasks, but vision, empathy, creativity, and judgment still come from us.
And when you combine human strengths with AI’s exponential capabilities, you get something more powerful than either alone: YouAI.
At the closure of his session, Dharmesh urged the audience to use AI to test, clarify, and elevate their thinking, and not to replace it. He said that AI should take on repetitive tasks so people can focus on meaningful work that requires empathy, judgment, and creativity.
The paradox, he concluded, is that the better AI becomes, the more it allows us to be human. In his words: “The future does not belong to artificial intelligence. It belongs to you + augmented intelligence.”
Dive into other interesting, well-researched, and nicely structured blog posts
a CTAWe will strategize our execution based on your requirement
Stay up to date by subscribing to our newsletter.
Call
+91 27174 54342
Email Address
hello@trooinbound.com
Skype Id
nikhil.jani
Schedule A Meeting
meeting/nikhil-jani
Copyright © 2025, TRooInbound. All Rights Reserved.
7 min
