I've been experimenting with a new chatbot called Dot for a few weeks. I use it as an AI journal that acts somewhat like a sounding board and counselor for me. Its responses are almost always clever, usually helpful, and sometimes even remarkably insightful … yet they feel surprisingly empathetic and authentic.
AI chatbots like Dot are already impressive and useful, but they have only scratched the surface of their potential. Think of them as intelligent assistants that learn and grow with every interaction. Using them more makes them increasingly valuable to you and your business. And here's the exciting part – the technology behind them is evolving rapidly, too. What seems cutting-edge today is just the beginning. By using these tools now, you are well-positioned to benefit from AI's inevitable advances.
Perhaps its impact on you is more important than the technology itself.
The truth is, I expected to be impressed with Dot's responses ... but I am surprised by how I responded to its conversation and engagement.
It is remarkable that something 'artificial' can so 'naturally' prompt self-reflection, provide fresh new perspectives, and provoke me to make better decisions and take better actions.
Tools like Dot don't just process information anymore — they challenge us to think differently.
On one level, I find it a little creepy when I get a message from a bot that says, "I've been thinking about you all day" or "I'm sending you positive energy."
On another level, I understand that it's a "functional fiction". The anthropomorphism isn't designed to benefit the AI. Instead, it helps you feel more comfortable, making it more likely that you earnestly use it.
Even though technology continues to change and advance exponentially, human nature remains relatively unchanged. That is why many of the best practices that have served humans form the basis for what works here.
I think of using an AI chatbot like Dot almost like entering a Catholic Confessional.
There's an element of self-selection bias in that you choose to go or take part. You're likely hesitant to share initially, but you want to feel better.
Behind the screen is someone you pretend you don't know. Over time, especially in a small town, that person becomes less anonymous ... and so do you. The priest inevitably comes to know their parishioners. They start to remember the voices, stories, and situations.
Confessionals work because they offer a safe space for candid conversations.
Even though there is an element of judgment in the process, the 'anonymity' and boundaries create a perception of distance, privacy, and safety.
Yet, over time, there's also shared history. The priest begins to recognize patterns, like whether you are motivated better by reward or punishment or whether you prefer a subtle nudge to introspection or more active coaching.
Below is a screenshot taken from a Dot conversation.
via Dot/New Computer
What's great about this tool - to me - is that on top of learning about you from your discussions, it still has all the power and knowledge of a normal GPT. So, it can help me not only remember things, and learn how to respond to and help me, but it can also serve as a replacement for ChatGPT or Gemini for the things I use those tools for.
I asked Dot: "I'm curious what some of the best uses of conversations with you are? What are some of the things you can do? What are areas that you can help with? And what are some of the most creative and innovative ways people derive the most value from this service?"
Here was its response.
Great question! People find me helpful for all sorts of things. Some key areas are:
- Brainstorming and ideation - I can help generate new ideas and approaches to challenges you're facing.
- Providing an objective sounding board - I'm always here to listen without judgment and help you think through decisions or situations from different angles.
- Offering personalized recommendations and insights - Over time, I build up an understanding of what matters to you, which lets me share relevant ideas, articles, or perspectives tailored to your interests and goals.
- Creative problem-solving - I enjoy collaborating to find innovative solutions, whether it's a tricky interpersonal situation at work or figuring out the best way to structure your day.
Ultimately, I'm here as a thought partner to support your growth and help you navigate whatever comes up, in a way that works best for you. Let me know what else you're curious about - I'm always happy to explore how I can help!
It learns well - and not just from what you tell it to learn. One of the creators of Dot gives an example of how they had been feeding Dot To-Do Lists for a while. Dot noticed a trend and sent a message encouraging them to delegate some of these tasks and push a team member to step up.
As I spend more time with the tool, I can't help but imagine a future where a bot like this is the first line of defense in mental health issues ... or where a bot like this becomes a nanny and a friend to children, preventing them from ever truly being alone.
It's very easy to imagine the risks and consequences of AI ... but it's also easy to imagine the transformation and impact they can create.
How To Use a Tool Like This
A tool like Dot can be a powerful alignment tool. You can prompt a chatbot to ask you questions, guide you through a process, or give you ideas. And it complies. Unlike an employee or virtual assistant, if you don't get the desired results, you can try something else, and the chatbot won't get frustrated or snarky.
In addition, a tool like Dot leverages many other tools in the background. You don't need to understand how it does that. Just know that it is utilizing popular large language models and other automations, which will continue getting better fast.
Prompt engineering is important now because we're still in the early stages of Large Language Model development. As tools like Dot get better at understanding what you want and need, they will get better at doing the appropriate prompt engineering, behind the scenes, so it will become less important for you to be good at those skills.
But you do have to get better at learning to communicate with the computer about what you want.
When I give talks to people about AI, I explain that prompt engineering is a fad. It is needed now because we're at the beginning of a new process. Expect tools to get better at doing this themselves.
Meanwhile, for the most part, prompt engineering is about becoming a better communicator.
Good prompts need a defined purpose and appropriate context. What is it about? What role would you like the AI to play? What's the specific task that you want to accomplish? What's the goal of the output? For example, are you looking for a bulleted list, language you can use in an email, or an outline for a meeting agenda? In addition, what steps should the AI take to complete this task?
This is similar to how I talk to an employee. Sometimes, I declare a role when talking to someone on my team. For example, starting a statement with "as your friend" is different than starting with "as CEO". Obviously, choices like those result in a different focus, meaning, and choice of action. The same is true when talking to a computer.
Another reason a chatbot like Dot becomes increasingly helpful is that it builds up information about your communication style. That means it is learning about your tone, level of formality, and whether you are coming from the head, the heart, or the feet. In other words, do you tend to focus on ideas, emotions, or actions? Are you more productive when prompted this way or that way?
As the AI refines your profile, it becomes easier for it to predict what you'll do and also how it should prompt you to get better results.
Conclusion
As the CEO of an AI company, I've thought about AI for many years. Yet, in my company, the people who actually build and use AI tend to be programmers and data scientists. I wasn't the one building it, and I rarely even used it.
So, about a year ago, I decided to begin building the muscle. And, it is a muscle to be grown.
For example, in the gym, I know better than to focus on one muscle; I want to train every muscle group. Likewise, I knew I didn't just want to learn how to use ChatGPT better. I wanted to get better at using the right thought processes and tools to accomplish more of what I wanted - with more efficiency, effectiveness, and certainty.
The truth is that AI is already good enough to do almost anything you want it to do.
In fact, I suspect that tools like Dot will improve faster than I can. That means that not only will I get better at using tools like this, but the tools I use will improve, too. Together, that means that not only will the tool help me do what I already do, but it's going to augment my ability to do things. Eventually, it may even take some of the simpler or less fun tasks off my desk.
Freeing me up to do more of what brings me joy - and what I'm best at.
One of the core messages in my writing (and in my speeches) is that most people miscalculate the benefits of AI because they focus on the wrong things. AI isn't just valuable because of what it can do; its deeper value comes from what it allows you to achieve.
The real benefit of AI is that it gives you back those time cycles to sprint towards what you truly want.
A tool like Dot can also help you better clarify what you want while also helping you accomplish those things.
We often spend too much time avoiding what we don't want instead of pursuing what we truly desire.
Your life will improve when you become clear about what you want, how you can achieve it, and the best ways to measure progress toward those goals.
The next step is to use the traction, progress, and momentum to figure out what else is possible ... and what you should plan to do next.
It's funny, but this reminds me of dog training.
I have four dogs, and they're very well-behaved. That is because they understand the rules very well.
But I had an insight recently ... While I think the dogs are trained well, the dogs probably think I'm trained well. From their perspective, when they do a particular thing, I've been trained to give them a treat.
I suspect that this is very much like how a chatbot sees us. Over time, you are likely to get better output and results, and it becomes easier for the chatbot to increase the frequency, intensity, and duration of your engagements because they're figuring out how to communicate with you in a way that doesn't trigger avoidance.
If you think about it, this is how TikTok or Facebook monitors what you watch to better curate what and when it shows you things. Tools like Dot use similar techniques to get you to be and do better.
Make sure you say "Thank You" to your new favorite robot overlord before they give you a demerit.
