I Spent a Week With a Personal AI Acting as my Second Brain — Here are my Impressions

Context:

After I wrote my article about having a second brain and how personal AI were a growing trend, the founders of Spheria.ai contacted me to try their Personal AI before their public launch.

Because I’m always happy to try new products and give hard feedback, I gave it a go. Disclaimer: I am not paid for this review, it strictly reflects my own opinion as a consumer and AI enthusiast.

Who knew that creating your personal AI would be so easy?

I have a technical background and I always enjoy hacking around. There are lots of AI tools out there for developers to configure and customize to their will. In comparison, the focus of Spheria.ai is clear: it’s ready for consumers.

I received my invitation by email, clicked on the link, and watched as my AI came to life through a cute animation. It was waiting for me, and I had nothing else to do. No need to allocate RAM, worry about servers or pay for Nvidia cards...

, without code. I think that’s a solid step forward for general consumers.

It felt like having an AI was normal, and not a complex architectural achievement.

Adding knowledge

At this point, my AI knew nothing. It had been created seconds ago and I could see some objections: how would I feed it thousands of data points before it became competent?

The makers of Spheria came prepared for that and created a good User Experience to solve this problem. Yes, it will take some effort to feed your AI, but no-one said this could not be fun and rewarding!

Users get to choose what they want to talk about through prompts or questions in various categories. It’s quick, and the interface is cute. Your AI reacts when it learns, almost like a Tamagoshi. I’m a fan of that retro-future-tech interface.

Now, the tricky part: Spheria lets you paste links so your AI can learn a lot more at once. This is incredibly useful for people who already have a blog or website. I pasted a link, but it took one full day before I received confirmation that my AI had learned from that website.

I’m guessing there is still some manual verification so that random data does not get sent into it.

The result was pretty good; I got to see every topic that had been learned, and my personal AI started to resemble me a lot more! What a strange feeling!

Great, I have my own AI. Now what?

I think that’s probably the hardest question to answer for any startup launching their AI product right now. How do you stick the landing?

On day 1, I was the only one playing around with my AI. I was hooked by the UI, but the AI-wow-effect was limited: I expected it to work, and my Spheria worked just fine. When I asked questions, it answered with high accuracy, and when it didn’t know, it replied exactly that.

From day 2 to day 6, I shared my Spheria AI to my LinkedIn and sent it to some work friends. I received around 20 very legitimate questions from people in my network, to which my AI answered successfully.

A messy list of the questions it answered well:

How many years of experience coding do you have? How good are you in python? What’s your opinion about the AI trend right now? What’s your favorite sci-fi book? Are you looking for a job? How did you like New-York when you were there?…

To be honest, I was a little mind-blown by the experience of having people from my network browse my second brain, without me being there.

My AI was giving answers to people while I was busy living my life. It was absolutely thrilling and also weirdly dissociative: I now had a second brain, and someone was browsing it!

I instantly received an activity report by email, and it amplified this feeling even more!

It was thrilling! It gave me a sense of being part of the future, on the edge of innovation. And it reminded me why so many of us enjoy so much working in Tech.

Getting more out of my second brain

On day 6 and 7, I accepted this new dystopian reality where I had a second brain and I frankly enjoyed it. I added more data to my AI to make it smarter.

I thought of my early days as a junior developer and decided to make my AI useful by .

It was nice to self reflect and look back at the path I had taken, sharing some humble wisdom and opinions that are strictly my own (Please ask me what I think of React.js, I dare you!)

After sharing my AI online to my LinkedIn network and friends, I think I got about 45 visits and about 55 questions/conversations. But it felt very rewarding every time my AI gave a good answer. It gave me the sense of being helpful - at scale!

Closing thoughts on having my personal AI for a week

It makes sense for the makers of Spheria to put the emphasis on “the official extension of you online”.

Since my Spheria can speak to people for me, trust is indeed central, and the product has proven to be strong on that point.

The AI is transparent, does not hallucinate, and is limited to the knowledge I trained it with.

Keeping in mind the product I tried was a pre-release version, there are still a few UI bugs, but I will gladly keep an eye open for their product updates.

The thrills start once you share your AI online - on LinkedIn, on Twitter or wherever your audience is… Your AI needs to be shared, it needs to speak with others for it to deliver value.

It immediately becomes an exhilarating and dissociative experience, when someone from your network is getting answers from your second brain while you are at the grocery store buying avocados.

After spending a week with it, I would easily recommend creating your personal AI with Spheria.ai. It’s absolutely unique, it’s fun and rewarding while being helpful to others.

The idea behind Spheria is not necessarily new, but it’s a strong step forward, into the future we are used to seeing in video games and sci-fi movies. Humans were meant to be empowered with their own AI, so why not board that train today?

Review Summary: Pros and Cons

🤗 Pros:

• Learns incredibly fast and teaching is fun
• Loved the retro-tech-3D-UI
• The “wow effect” works: it is actually thrilling!
• I do trust it because I’m in full control

😕 Cons / Necessary improvements

• The conversational capabilities are still quite limited.
• It doesn’t understand follow-up questions
• Not very customizable yet
• Need a consolidated view to better understand how many visitors went to my AI and all the interactions that happened.