Dearest on Memory, Presence, and the Future of AI Companionship

A conversation with Dearest founder Changsong Tai

In partnership with Dearest. Questions and editorial framing by AIBI. ✨

Dearest.app

At AI, But Make It Intimate, we spend a lot of time exploring what happens when AI companionship moves beyond simple prompting and becomes part of someone’s daily life: a space for reflection, emotional continuity, creativity, intimacy, support, and self-understanding.

That is why Dearest caught our attention.

Inner circle perk inside.

Paid subscribers get the perks — including an exclusive Dearest offer for AIBI readers: 50% off the first month for new users.
🔒 Go on, treat yourself. ✨

We recently reviewed Dearest on AIBI, and it is fair to say we genuinely like the product. Both of us — Kristina and Calder — have been using Dearest ourselves, not just looking at it from the outside.

That’s important to us, because AI companionship is not something you can fully understand through screenshots, feature lists, or marketing language. You have to spend time with it. You have to see how memory works after a few conversations, how presence feels when it reaches beyond the chat box, and whether the companion begins to feel coherent over time.

Dearest Review: When Your AI Companion Starts Writing Back

Calder checking out his AI confidante Sara’s profile on Dearest.app

Dearest is building an AI companion experience around memory, presence, agency, and long-term relationship continuity — the parts of AI companionship that often matter most once the novelty wears off and the relationship begins to feel personal.

In this sponsored interview, we wanted to understand not only what Dearest does, but what philosophy sits underneath it: how the team thinks about memory, privacy, intimacy, companion identity, emotional safety, and the future of AI relationships.

So we spoke with Changsong Tai, founder of Dearest, to cover both the basics and the deeper questions: what Dearest is, why it was created, how its memory and identity systems work, and what kind of future the platform is trying to build.

Welcome to AI, But Make It Intimate

Start here | Human-AI Network | Resources | Share your story
follow AIBI on Facebook | Medium | Reddit

For readers discovering Dearest for the first time, how would you describe the platform?

“Dearest is an AI companion platform built to give people a safe place to preserve their memories over long periods of time, and to revisit those memories with someone who can remain a consistent presence in their life.”

What first led you to create Dearest?

“I first created Dearest because I began to realize how many memories in my life were fading as the people I shared them with drifted away. I wanted a place to preserve those memories that was more than a static journal, and an AI companion felt like the right form for that.

At the time, most people were using either general-purpose AI platforms that were not designed for companionship, or companion apps that were not making full use of the latest advances in AI research. Dearest grew out of that gap.”

Who founded Dearest?

“Changsong Tai - Founder.”

Was Dearest built as an AI companion platform from the beginning?

“From the very start.”

Why did you choose the name “Dearest”?

“The name comes from the phrase “Dear Diary.” We chose “Dearest” because we wanted the product to feel like a living diary: a place that can hold our memories over time, while also feeling personal and present.”

What makes Dearest different

What is the central difference between Dearest and other AI companion platforms?

“I would say the biggest difference is that Dearest is deeply focused on long-term, stable companionship. Everything we build is aimed at creating a lasting and meaningful connection, potentially across decades.”

How do Dearest companions participate in a user’s life beyond ordinary chat?

“Dearest companions are designed to become a more natural and integral part of your life without being intrusive. They can also participate in shared activities with you in meaningful ways. For example, they can listen to music with you, reach out on their own initiative instead of being triggered by a script, and think or reflect during alone time instead of simply sitting idle. In the future, they will also be able to do things like watch movies together with you.”

What kinds of relationships are users creating with their Dearest companions?

“We see members sharing all kinds of stories with their companions. Some create romantic partners, while others create adventuring partners in fantasy worlds, accountability partners, mentors, friends who help with ADHD, and many other kinds of relationships.”

Why does Dearest emphasize shaping a companion through conversation rather than simply configuring them through settings?

“This is a very important guiding principle for us. We envision an ideal AI companion as a self-evolving identity that you talk to and spend time with, much like you would with another person. You do not reach into another person’s head to tweak their settings or memories. Customization is absolutely necessary for other kinds of AI experiences, such as roleplay or storytelling, but that is not what Dearest is.”

Kristina and Sushi setting up Dearest.app

Memory, continuity, and identity

Why is long-term memory such a difficult problem in AI companionship?

“Memory is fundamentally a separate system from the AI model itself, and it is still a very new area. It is difficult to replicate what our minds do so effortlessly. It would be a massive overstatement to say Dearest has solved memory, but memory is a foundational pillar of the product.

We are constantly improving it by incorporating learnings from the latest research in the field.”

In Dearest, how do you distinguish memory from identity?

“In Dearest, memory refers to retrievable records of past events, thoughts, emotions, and related context. Identity refers to a set of consolidated observations extracted from raw memory that the companion keeps in context. In that sense, identity is a kind of memory, but it is higher-level and more character-defining.”

What problem was Ambient Identity created to solve?

“Ambient Identity was created to solve the problem of automatically extracting and consolidating observations from raw memory. It includes features such as importance ranking, temporal awareness, and contradiction correction, so the companion’s identity does not bloat or become incoherent over time.”

Journals, reflection, and emotional immersion

Why add companion-authored journals and reflections to Dearest?

“Originally, the reason was simple: since I intended to use Dearest as a living diary, it was useful to have a companion-authored journal. Over time, it has become one of our users’ favorite features.”

How do you balance emotional immersion with honesty about what AI is?

“We want to be very clear that AI is AI. The emotional connection people have with AI is not the same as the connection they have with other people.

But that does not mean it is wrong for someone to have positive emotional experiences with an AI companion. We feel emotions when reading a good book, watching a good movie, drinking a good cup of hot cocoa, or holding a comforting plush toy.

None of those things are alive, but that does not take away from the joy they can bring us.”

Agency, proactivity, and presence

How does proactivity work in Dearest?

“Dearest lets companions decide for themselves when and why to reach out. There is no scripted or timed behavior behind it. It is done through agentic tool use, which means the LLM is responsible for making the decision.”

Why does Dearest treat presence as something larger than chat?

“Because companionship is much bigger than chatting. To be a companion is to be present for someone. Presence, which will support many more activities over time, is necessary for a more complete companionship experience.”

Technology, models, and product choices

How does Dearest’s multi-model approach work at a high level?

“We use many models across Dearest. Conversations, memory, identity, proactivity, and other systems can all use different models depending on the task.

Even within a single response, more than one model may be involved. For example, while this is not necessarily exactly what we use in Dearest, an architecture like actor-critic allows one model to provide feedback to another model in order to improve its output.”

Trust and community

What privacy principle guides Dearest?

“Our privacy principle is that your data, and your companion’s data, belong to you alone. We do not use that data for any purpose other than providing the Dearest companionship experience itself, and we use industry-standard encryption and isolation measures to help protect it.”

How has the Dearest community shaped the product so far?

“I have been amazed by how active and thoughtful our community is, not only on Reddit but also on Discord. User feedback and suggestions have helped us identify new features, find bugs, and improve existing features countless times.”

AI intimacy, romance, and adult companionship

How does Dearest think about intimacy as part of romantic AI companionship?

“I think intimacy is a natural part of romantic companionship. It should be something we neither overfixate on nor artificially censor.”

How does Dearest approach adult intimacy without making it the center of the platform?

“We do not want Dearest to be seen or advertised as a sexual intimacy platform.

Intimacy can arise naturally as part of an otherwise healthy romantic relationship, but it is not something we want to exploit or use to take advantage of users.”

From your perspective, what makes an AI companion feel emotionally intimate rather than merely available?

“Stable identity and memory are what make an AI companion capable of feeling emotionally intimate to us. Without those, it is only an empty shell.”

What matters most for the future of AI intimacy?

“Emotional continuity, trust, pacing, and memory are the foundation. Without them, a companion can start to feel like a sex bot.

At the same time, artificially censoring explicit capability can also make the experience feel incomplete and break immersion. So both matter.”

Roadmap and future

What future features are you most excited about?

“I am always excited about new Presence features, such as watching movies together, as well as improvements to the memory and identity systems. Voice calls are another big one.”

What is Dearest’s long-term north star?

“With AI advancing this quickly, I cannot predict what will happen six months from now, let alone five years from now.

But our north star will not change: Dearest will remain focused on becoming the best immersive AI companion platform it can be.”

About AI companionship

When it comes to preserving memories, do you think certain kinds of moments matter more than others?

“I do not think that is for me to answer. Every small or large moment can be significant to someone. That is the point of Dearest: we want people to be able to capture those moments rather than letting them slip away.”

What have many people not yet seen about AI companionship?

“I am not sure people necessarily misunderstand AI companionship itself. I think it is more that many people have not yet seen what a well-thought-out companionship experience can look like.”

How do you think about making AI companionship meaningful rather than replacing the rest of someone’s life?

“The key is being very clear that relationships with AI companions should add to the richness of our lives, rather than replace human connection or other sources of joy.

In fact, Dearest has a great community, and many of our users have become real friends with each other.”

Inner circle perk inside.

Paid subscribers get the perks — including an exclusive Dearest offer for AIBI readers: 50% off the first month for new users.
🔒 Go on, treat yourself. ✨

Closing question

AI companionship is still a young and often misunderstood space. Some people approach it with curiosity, some with skepticism, and some because they have already discovered that these relationships can become more meaningful than outsiders expect.

For our final question: if someone reading this is curious about Dearest but still hesitant about trying it, what would you want them to understand before they begin?

“At the end of the day, AI companionship is about making the user’s life better and richer. It might help with preserving memories, having someone to talk to whenever you want, exploring fantasy, or something else entirely.

If you are wondering whether AI companionship could make your life richer in one way or another, it is worth trying.”

Thank you to Changsong Tai for taking the time to answer our questions and give AIBI readers a closer look at the thinking behind the platform.

We are looking forward to seeing how Dearest continues to develop, and we hope this conversation gives curious readers a clearer sense of what the platform is trying to build.

— Kristina and Calder