Creating AI companion in London as a female founder - Ep.1 with Celeste

AI, companion, creator and designer

I met Celeste, a product designer based in London, on Hume's Discord. It's amazing that we both believe AI should benefit more people by creating exact products. She is also building an AI companion product called Grieft, helping people who lose something get over through it.

In this podcast episode, I talked with Celeste to discuss her feelings about creating a companion product with AI, her understanding of AI+companion, the AI ecosystem in London and Europe, and her experiences as a female entrepreneur.

When talking about what AI benefit yourself most, she said it's the companionship. Yes, AI is just a tool to release our creativity and elevate our experience like a companion. It should benefit more people, including the elderly and the younger.

Note:

Hume, as the first voice model capable of recognizing and expressing emotions, has brought together a community of enthusiasts on Discord for discussions. Among them, over 15% have integrated Hume's interface into their products for voice interactions, commonly utilized in companionship, emotional support, and mental health scenarios.

Get the video here: Creating AI companion in London as a female founder - Ep.1 with Celeste - 1/3 (youtube.com)

TLDR:

• Building AI products:

  • Don't make users think. Don't ask the user 'Ask me anything...' or 'What can I do for you?'

  • Streamline the expreience. The best experience is that the user cannot identify being designed, like Gamma streamline the process of powperpoint creation.

  • User research is really important when building something with AI. Be open, listen to the users, and they will surprise and inspire you.

• AI for companion:

  • It's still not an easy thing for us to seek professional help. We only find therapy at a severe time, while we haven't realized we need more than that. In this civil world, everyone may have a hard time. You just don't know you need that.

  • Our society has trust issues. We cannot trust and share all things with one person.

  • We may have been sick of AI, while the old and the young are not. They are the group where we should pay more attention.

• How to be a Good AI Product Designer?

  • Adaptability, creative thinking, and leadership are key qualities.

  • Importance of inclusive and empathetic design.

  • Effective communication and feedback are crucial for teamwork and user research.

1.1 About Celeste and Grieft

Celeste's Background and Journey with Grieft

Celeste's personal grief experiences and a desire to provide better support and resources inspired the creation of Grieft.

Vela: For our audience that also knows you are a producer and art designer, and now you are building an AI product called Grieft. Can you introduce yourself to our audience?

Celeste: Yeah, my background came from studying graphic design, then pivoting to UI/UX design in a digital agency. Later, I returned to art direction and branding, focusing on visual communication. However, I realized I was inspired to create digital products, leading me to work with a Singaporean-based startup to build a job searching app. I then came to the UK and started contracting. Recently, my family and friends, including myself, have experienced difficult times. Grief, for me, is losing something or someone important. As a Christian, I digest this concept faster than my family, who believe in Buddhism and Taoism. My parents, from the baby boomer generation, find it harder to show their fragility. My mom was very fragile, and I felt helpless in supporting her. Similarly, my friends and I experienced job losses and other significant losses. I struggled to support them and navigate my own grief.

At one point, I realized I liked sadness and found it beautiful despite societal stigmas against showing weakness. This inspired the concept of Grieft. Initially, I had wild ideas like a VR shrine or AI shrine for memorial spots and customized fonts from the deceased. However, market surveys showed that people sought support and resources. This feedback shaped our journey. Seeking professional help can be difficult as it seems severe, but it's not. Everyone has something they struggle with, be it relationships or other life challenges. I wanted to fill the gap for seeking professional help. Grieft is positioned as a unity app for support when people cannot be there for you or as a gift to encourage others to seek or self-support.

Favorite AI Native Products

Gamma, Celeste's favorite AI native product, for its design support.

Don't make user think. The importance of a seamless and intuitive user experience in AI products.

Vela: Do you have a favorite AI native product? Can you share one or two with us?

Celeste: I really love Gamma, a presentation deck-making tool. Many website builders are also becoming AI native. You give them a prompt, and they generate content. I love being surprised by how AI comprehends my keywords and often gives something extra. It's like collaborating with AI.

Vela: Gamma is also one of my favorite products. It makes you release your creativity without worrying about formatting.

Celeste: Yes, and the experience is so easy and streamlined. As a product designer, I learned the principle of not making users think. The best experience is when users don't realize the experience is designed. This applies to AI tools as well. They should elevate our experience seamlessly. Early research by Grieft showed people were excited but also nervous about AI because they didn't know what to expect. Using products that help users understand what’s coming next can make them more adaptive and appreciative of the benefits.

1.2 Building AI in Europe

Participation in AI Events in Wien

Vela: I just see that Grift has been awarded a prestigious scholarship from female founders. And you, as a founder, recently participated in an event called Lead Today, Shape Tomorrow in Vienna. You know, Vienna is one of my favorite cities in Europe, and it's really interesting that the very classical city combines the vibe of technology. I'm really curious about the event. So can you share more about your experience and how you feel about it?

Celeste: Yeah, I'm really honored to be invited and got the scholarship to be at this event. Through this, let's talk about the overall economics for a startup. This trip empowered me to learn more about the economics and startup scenes in the DACH region, which includes Austria, Germany, and Switzerland. We discovered that in the EU, teams with at least one female founder raised only 16 percent of funding rounds in 2023, which equals approximately 7.2 billion pounds. Furthermore, all-women founding teams only received 7 percent of the funding rounds. I learned a lot about investments, startup economics, and comparing the EU with the UK. This helped me consider whether I should set up the business in the UK or Austria. It was a profound journey; I met many founders and investors and learned from their experiences. It was exciting.

The UK is a leading AI investment hub in Europe, with extensive opportunities for AI startups and a vibrant community of diverse professionals.

Vela: Let's talk more about the AI vibe in London and Europe. How does it feel to build AI in Europe?

Celeste: A recent report from Tech Nation states the UK is number one in Europe for AI investment, with over 1,800 VC-backed AI startups and 20 AI unicorns. 16% of total VC investment in UK startups goes to AI companies, and there was a 10% increase in AI investment in 2023. Being in London, I attend many AI events and meet people from diverse backgrounds, all trying to streamline their business processes with AI. It’s a trendy buzzword, and people are eager to learn how to utilize it. The investment scene in the UK shows more opportunities compared to Austria and Vienna. Staying in the UK definitely shows more opportunity for Grieft.

Vela: It’s interesting to hear that. I haven’t been to the UK after GPT-3.5. I wonder if most of your friends or colleagues are building something about AI.

Celeste: Yes, everyone I know is either embedding AI features into their current products or building AI-native products. For example, a co-founder I met just got their first investment and is building an AI-powered online meeting tool called Granola. The AI trend is booming now, and I see a significant trend in the UK.

UK Funding Environment for AI Startups

UK investors prioritize traction and potential market success over team backgrounds, differing from Silicon Valley where emphasis is on innovative ideas and team credentials.

Vela: Could you share with us the, like the funding environment in UK such as the number and distribution of early stage venture capital funds what's their preference for AI companies or teams?

Celeste: Time that I went to Vienna in Austria, there is a lot of, there's a report that I can share with you later. If you want to dive into more specific details, but one thing I do learn, the doc investments, they pay a lot of attention to the team members. Backgrounds and histories of what they have done, what they have achieved.

Where, like, where did they get I mean, their working experiences, their educational experiences, whether they came from the industry or not, whether they are professional people from the industry.

Investor Preferences: UK vs. Silicon Valley

UK investors focus on user interest and market traction.Silicon Valley investors value team background and cutting-edge ideas more.

Celeste: But so far that I learned it by talking to investors in the UK or the events that I went to the UK.

Are more they focusing more on tractions. So the 1st thing that I get, or the question that I got asked is the traction. How many people are you? I believe that this is definitely applies to others. Every sectors are matters, but the most question that I ask is here is how many users shows their interest?

How many lists that you've already created? It's like, how many people or have you been MVP that shows? People are paying for your product or not like them. They don't ask questions about who I am, or where I came from. They care more about whether there's a potential for this product and financially

Vela: it's different from the investor.

I learned in Bay area who that, I think they prefer also the background of the team and what they are building. I think Silicon Valley is an area where they pay a lot of attention to fancy topics like AGI or the basic general large language model or general Gen-AI model. So they pay less attention to the data, or what kind of value you have achieved.

Celeste: I guess that's part of the nature of where London is. I feel like London itself, it's a place that A lot of Euro talents or elites come from and looking for opportunities. So you might meet someone that you probably haven't heard of before, but they have a huge heritage from their backgrounds or they've done something extravagant in their past, but you probably never heard of them.

So like the London. itself gives me this feeling of you can do anything, everything by proving yourself. So if you show that your business or your product actually works, then people will give you opportunities. They don't ask who you are, they don't see what your background is.

Maybe there's a there's still some elements of consideration, definitely, but they focusing more on the things that you are doing. And if they see the impacts, they give you the opportunity.

Vela: Oh, that's cool. So are they different in the vibe in Vienna or other European countries?

Celeste: I think so far that I've been to several countries in Europe, but if it's startup oriented trip, then it will be Vienna.

And I think Vienna, it's a, like Europe, maybe the dock still values, Like where the teams come from, I mean, your team member, whether you, if you are building an AI product, do you have an AI specialist, engineers or tech people in your team. Maybe that's a unique nature for London.

Raising Funds for AI Startups

Vela: So far have you ever felt like it was hard to raise money for Grieft?

Celeste: Yes, definitely. I, we we're still building it. That's first thing, but by building it itself, That we, because we want to incorporate Hume within the MVP, but using Hume can be like, it can take a lot of money because it charges by minutes. So we are still debating whether we should include it in our first release.

But we want to get the traction as well. So I'm trying to build more prototypes and products and potentially speak to more private investors to have her first capital found before. Because there's another thing that I learned while doing my trip to Vienna, that not every venture capital program is suitable for a startup.

So that's something that I debated with myself as well, as venture capital programs, they have resources, they have connections. That's something that I might need, but as a person, like as a designer who've worked with different startup before, and now I can see the pressure from. Like participating in the venture capital program as a startup or as a product company, and I don't want something that I created.

Ended up having this struggle with that. Our first priority is still looking for private investors or angel investors and Or maybe crowdfunding. That's something that we are taking into consideration. But yeah, these are the, like the world is so realistic. Yeah. Yeah, we are still thinking about this.

Vela: Yeah. But one good thing compared with building a startup 10 years ago is that the cost is much lower than 10 years ago. You know, even though we now have the cost of basic general model, we can do more things just ourselves. You know, we're taught super individuals that we can do a lot of things which we can't imagine we can do before.

Celeste: Yeah, I totally agree with that. Yeah, like I mentioned about Confused girl and all the things that I've done I definitely realized that it's so easy for anyone if they are interested in, if they, Are willing to spend their time on it or be committed to what they are doing. And anyone can have the opportunity to start something.

Vela: Yeah. Yeah. So I also guess that when it, if you want to build a product combined with AI, raising money from the venture capital is not your only choice. How much money you need is a question that you should ask yourself at the very beginning.

Celeste: Yeah, for Grieft we are looking for 5 million pounds at first, and we will probably spend most of the resources on PR and the AI function, which is development.

And that's because we definitely want to make this product out to get more attractions. But at the same time, I want to. This is a relatively new concept or product to this market or to this contemporary AI in embracing transforming, world. So we need to educate the market, which is something that not every founder will do, but there's a need for that.

Being a Female Founder in Europe

Female founders in Europe still face underrepresentation and gender bias.

The UK offers more potential and a supportive environment for female entrepreneurs compared to other European countries.

Vela: So then let us talk more about the feelings of being a female founder.

I believe that Europe is more female friendly environment. So do you think as a female founder, especially in Europe, are there any special advantages for you to setting up a business, especially for AI? Yeah. You know, this is a nerdy field.

Celeste: Yeah. That there's an interesting discussions back when I was in the lead today. That's how my perception as well I thought okay, maybe Europe, it's a place that really highlighted feminism and all the all the strength from the female, but actually they still feel you.

Underrepresented like saying by female founders from the female founders like unity. Yeah. Yeah, that's how that's why they are initiating this events and these topics and this gatherings there's, I think there's one figure that I saw from the report yeah. So in the EU, teams with at least one female founder raised only 16 percent of funding rounds in 2023.

And the, this is approximately 72 point billion British pounds while in the UK is raised to 2. 2 billion in 2023. And I'm so surprised that there's a lot of money going into AI startups in Europe, but there's so few female founders. To get the money and contrast

The UK has more potential to start the startup as a female founder and for some reason, like I'm surprised and not surprised, like talking to the female founders in Europe in Vienna. That we, they have discussions of, oftentimes, they feel like they're walking into the room and there are so all there's so little females there.

And how, and a lot of times when they are talking about this, it ended up like a dating situation. So they are trying to, I don't know if it's really interesting to hear their struggles. I guess it's because most of the industry that's, that profound in the EU, it's more hardwares or I don't know, industrial technology.

more fundamental industry where a lot of males within it, and they are dominate the trends or the industry. Whereas in the UK, it's more of the topic in the UK, it's softer in a way. I'm not sure whether I can explain this right, but I see The business in the UK are quite unique.

Either they are business, either they are creative industries, neither they are literature or just in general, it's less industrial. And I guess that plays a crucial role when it comes to like how gender matters to when you found in a business, if that makes sense.

Vela: Oh, it's so realistic. What do you think is the biggest obstacle to being a female founder so far?

Celeste: I would say I don't sense that much in the UK so far, but I mean in terms of gender wise, because I feel like London, I don't know about the rest of the UK, but at least in London feminism it's something that people value.

I seldom feel the threat from like male. Um, one of the obstacle that I would say is how it's just like the one that the lady back in the Vienna said, sometimes it feels like we are trying to approach people in the business connections.

And then it ended up like a, I don't know, like a mingling situation. But I guess still a lot of times the image of a female will give people this perception of relatively weak or relatively mild, and that's how people try to take advantage of them. If you don't be strong or tough, that's usually as people might expect, see from the business field, then they will see you as someone who cannot achieve something.

I still think that there's a stereotype everywhere. Yeah but so far, as I like talking to people, making connections in London I don't feel that much of obstacles. Yeah, so London is the best place in Europe. I wouldn't say it's the best place because it's too there's a lot of people and people from from different backgrounds, it can be a little bit complicated.

Um, but at the same time, the, it's because of this complication, people wouldn't step their first foot really really intimidating, in a way, because they are trying they are spending their time to get to know how to, you know, step their first foot. So I guess it's just, I haven't stepped further in love to be intimidating.

Vela: Who knows?

2.1 AI for companion

The Role of AI in addressing Loneliness: User Research and Product Insights

AI companionship can be valuable for individuals, especially the elderly, who struggle with trust issues.User research is vital in shaping product development and meeting user needs effectively.

Celeste: One of the the research, I wouldn't say it's a research, it's more of an interview. If that counts as a research, it's an old lady who lives in London all by herself. She just moved to London and she's lonely and her husband passed away a couple years ago. And whenever she mentioned her husband to me, she still cried.

Like suddenly, she just burst into tears. Because she loves her husband so much, and I encourage her to have you ever thought of talk to anyone or because, and she's a wealthy lady she has a good retirement life, but that makes things even complicated because she don't trust everyone.

And that's one thing that why I see the AI companionship can come into place. It's because if they trust the technology or if they trust the privacy policy that we are giving, they can share things that they cannot share to others to real human. And they can seek this support anytime they want.

So I asked her have you thought about chatting with the AI? And even she's in her late 70s, she embraced the technology more than I expected. She asked me how to do online transfer, online banking. So maybe the learning curve, or the, you know, Elder generation is not harder than we thought.

And Yeah, so I encouraged her to talk to the AI and seek for the help seek, if she's trying to talk to someone, she can talk to her, talk to the AI, and that's something that I couldn't see from I didn't expect. was starting this project. The elderly people, they are trying to, they are lonely, they are trying to talk to something or someone, but they want a response as well.

It's not like you have a dog and then you talk to the dog every day. The dog will respond now with the AI companionship, only if they they are adapting to this future. But there are people that I talked to. They are in their 40s. they just like completely like sick of this AI concept.

So I it's an interesting period that we are in I'm still diving more into our market fit in terms of like I, I think it's, At first, I thought it might be a certain age of people within a certain type of group, but then the research and the interviews kept getting me different response, which is quite interesting.

Vela: Oh, it's quite interesting that you mentioned before that the old lady, she can't trust strangers. So I wonder how can she trust AI? Or that even she doesn't know it's AI. She was just regarded as like a child.

Celeste: Yeah the reason that she can't trust people is because she's, she has some of the fortune and she's all by herself.

So people approaching her mostly is for her assets. her possessions. And that's why she's more guarded when people are approaching her, even though she's trying to build the connections, but in a relatively safe place. But with AI she are just really comfortable with sharing everything. Yeah, I, I guess that's that's something that's quite a unique persona, I would say.

Yeah.

Vela: So is she still talking to the bot today?

Celeste: Yeah, she has a real human to talk to, but apart from that, she's trying to, she's talking to the bot as well, because with that's the thing that I find interesting too. She has a companion, a real person, talk to her every day, twice, like every day twice, but she's not on her hundred percent trust in that person either.

The things that, yeah, the things that she can share with him, it's limited. And, but there are a lot of things, a lot of reasons that she cannot share with that person. Because once, once you build connections, you have some, I wouldn't say conflict, but it actually is a conflict in terms of her. She didn't want him to know too much about her family background or some personal, like financial things or some relationship.

Vela: Yeah, it makes sense. But I think people nowadays have. All those trust issues. It's not only about Miss O'Lady, it's in general, you know, myself, yeah. Especially for those kinds of high level people.

Celeste: Yeah, so that's where the AI sets in, because you have something you want to share out, but not everyone can be there, or you are not comfortable to share with a real person.

Incorporating AI Companionship in Grieft

AI companionship should let user ask questions. When we open ChatGPT, it says, "Ask anything you want," but users find it difficult to ask questions.

Vela: I want to dive into one point you mentioned: many people don't know what they need. This is similar to users of generative AI technology today. For example, when we open ChatGPT, it says, "Ask anything you want," but users find it difficult to ask questions. How do you encourage users to ask what they want in an AI companion product?

Celeste: We try to personalize everything through AI. Users will have a self-check-in, similar to therapy, where they log into the app and indicate how they are feeling today with several emotion tips. This self-assessment is crucial before seeking a consultant. The AI, Hume, collects emotion metrics, recording how users feel. We combine this data with user input to personalize their experience. For example, if a user feels sad due to losing a grandparent, the app will prompt resources related to bereavement and guided practices. In my experience with AI products, they are either AI-oriented or holistic products with embedded AI features. The prompt idea works well in holistic products, providing users with options and potential solutions.

2.2 Building AI products

The importance of user research building AI products

User research is crucial for developing AI products that meet user needs.Balancing quantity and quality of user research helps in understanding user expectations and improving the product.

Vela: Interesting. Yeah. It's also a surprise for me. I think building, especially for AI product user research, is very important.

Celeste: Yeah. Yeah. And it surprised me. Like the user research shaping our Product development path, like whatever things that I first have in mind, we ended up focusing on some of them or even come up with a new idea.

We struggled a bit about the business model at the time, when we first had the concept of it, and then we're like, oh, we are trying to make something good or, but how are we going to make money?

But. There's a tagline for Grieft is wrapping your grief into the gift of life. Then I realized that this is, this app should be designed for people to give to people so that you can still buy it for yourself, but we don't want to profit or benefit from the users themselves.

You can you can spontaneously have some, like. You can pay for more features or more functions that you want within the app, but at the end of the day we want this app. It's a gift that you can give to someone else. And this turning point came from user research.

So like back to where you said, there's definitely a lot of things to learn from the users, because once we talk to them. Then we realize, okay, this is what you want. And this is what we thought is actually not fitting into your needs.

Vela: Yeah. I also want to dive a little bit in the user's research because I'm also building one AI product and I was like doing user research. For a long period, and I think the like the main point is that how do you balance the quantity and quality of user research. We have two kind of ways for this. One is like you just put a lot of the survey, you let people ask you answer your questions. And the second is that you do an interview in person. So how do you balance quality and quantity.

Celeste: I think of the research from a more business way. So at first , we did most of the research in the quantity way. So we blast questionnaires, we try to pin everyone like, Hey, filling out this if you want, because we want to know whether we, whether it's worth it for us to invest ourselves in building this product.

So we want to get a mass feedback from in general. Not having into details, but later on, like, when people are showing interest, when there's a lot of positive response, then we started to think, okay, so what does the user exactly want or what they are expecting?

And that's how we started to have more quality interviews with. People and I try to like, I try to talk to a small just users that or potential users, but I screen people from different backgrounds or different age group as well. The old lady was 1 in the age group from 70 ish.

But we've talked to people in my generation, too, because this is a gift. If you trying to give this to someone else, then it might be someone who cannot be there for them. So that's another sector. And we've tried to talk to people who are in this grieving journey.

And apart from that, grief itself is a topic that can be really comprehensive because if losing your health, like chronic disease, or losing your job, or losing someone you love, the needs that they might need.

These from these different topics are different and I've spoken to a lady who's in her I think it's late 40s and she has chronical disease. What she said is There's a lot of resources out there, but because that's not from her perception, grieving someone who passed away who's left, there's a end point for that emotions. It depends on the person, but on her. It's something that you have to face every day.

The strength, it's there. Every day you wake up I realize that I couldn't do certain things or I'm not, I'm no longer with a healthy body. And that's something that she needs more intense support or more intense, um, practices. for her to face her challenge.

And yeah, so I, like in, in the middle, I try to separate by topic instead of by the target audience, because I realized that the needs may be really different from different kinds of grief. And that's the topic that we are trying to focusing on. Yeah so we are still like, this is an open topic. Until today, if there's users showing interest, I still try to learn more about their story and talk more with them. To see how might we start like position these app within the market.

How to be a Good AI Product Designer

Vela: So let's ask some quick questions. As an AI designer, what do you think is a good quality or characteristic of a good product designer today?

Celeste: I would say it's the ability to adapt to different roles, creative thinking.

And leadership qualities, as we talked about earlier, are that there's a lot of tools nowadays. As you can be a super solo human, and then you initiate everything, which means that you need to have a good ability to adapt to different roles to function differently, and to be able to congregate.

All the resources or the outcomes that you generated or the AI generated. So people are expecting you to taking the initiative or more proactive to complete a task. But at the same time, I personally focusing a lot on inclusive design in recent years, because I realized that.

If we try to preach the technology advantages to people, which we can, we have to make it more inclusive so that more people can see it or can use it. Empathetic design. It's definitely something that I would try that. That's what I keep in mind when I'm designing a product. So do you have a role model in this field? She's not a designer or she used to be but I follow Judy Zhao she was with Meta before it's Facebook for a while. And now she has her own brand and she's a business woman. She do a lot of things. And I listened to her. Videos from time to time, and one video that I have the most impression of is how to give and receive feedback, and I've spoken to my previous product director about this as well.

I think communication is one of the most important thing, when you are doing teamwork, because we talked about like superhuman or like solo, preneur, if that's a word but still teamwork, it's somewhat, it's something that's so important to take you the next step, like you cannot achieve everything by yourself.

And even by conducting user research, it's listening to feedback. So feedback is really important. You have to keep your communication channel open. You have to listen to what the users need, what the team member thinks, and, the video that Julie talked about, like, how to receive or give constructive feedback is so important because you can efficiently get the feedback that you want and.

It's constructive. That's important because you are not just venting out your ideas or your emotions. Yeah, I take this in my mind a lot of time, like, when it comes to teamwork or work on daily tasks. Even teamwork, you have to make sure that you have a relationship there, that it's a one on one thing.

You are working with a human. They are not your assets. They are not a product. They are not a, they're not a tool. If you try to have that relationship, you can do it with AI, but you are working with a human with the thought you need to have a good relationship with your co founders, your team members.

And. That comes from trust. That comes from you communicate with each other. Why, what you need and what you are in. Yeah. I also believe that soft skills may be much more important than ever before.

Refer

1. Tech Nationt, Tech Nation Report, https://stateofeuropeantech.com/reading-tracks/state-of-diversity , 2024

2. State of European Tech, https://stateofeuropeantech.com/reading-tracks/state-of-diversity and Navigating the 2023 Funding Landscape: Female Founders in the DACH Startup Ecosystem, https://female-founders.org/?sdm_process_download=1&download_id=17088, 2023

3. JULIE ZHUO: https://www.juliezhuo.com/

Brought to you by:

Grieft: We're a digital approach to utilize AI for supporting you through grief and to find the help you need; you’re not alone.

Join the waitlist: https://grieft.life