Day 380 of Sprited

SpriteDX

Last year, I worked mainly on spritedx.com. It's been really fun product to work on and I intend to continue working on it if there is any inbound signals but for now it is in "optionality preserved" territory. Earlier this year, I put that project in permanent early access until I see any user requests.

Machi

Machi is a project I started last year and it is to give AI agents a place to live and interact. This is the core vision of my solo startup. I want to work on this further. Right now a tech demo can be seen in machi.sprited.ai.

Digital Being, Pixel

Pixel is the first digital being (a living agent) inside Machi. I'm honestly struggling with this. The main problem is decision paralysis.

• What does it look like?
• What can it do?
• Can it grow?
• Can it learn?
• Can it talk?

There are bunch of questions like these I need to answer and the most difficult part is:

• How do I make it lovable but still classy?
• How do I make it such that it is acceptable by society?
• How do I make it such that people come back to see it?
• How do I make it viral?

So what today?

These few weeks has been challenging due to lots of meetings and helping out with friends projects, so I'm working on getting the regular work rhythm back for Sprited.

Goal is to somehow get 8 hours of work hour. My kid pickup is at 3pm so I will need to start my day at 7am.

To do that I need to sleep early and wake up at 6am. And every night, I will prepare the food for next day.

// Summer Schedule 
6 AM: Wake up, go to co-working space, do course work (1)
7 AM: Do planning for the day. Signal & Noise analysis (2)
8 AM: Kid drop-off
9 AM: Goto office, co-working space, cafe, etc (3)
10 AM: Work (4)
11 AM: Work (5)
12 PM: Food (6)
1 PM: Work (7)
2 PM: Work (8)
3 PM: Kid Pickup (end of work), family time afterwards.

So what will you do today?

I will write this part in Korean for no reason. Nobody reads this anyways 🐶. It is also that I want to get better at expressing ideas in Korean.

지금 프로젝트 디렉션 중에 마치와 화이트룸 중에 하나를 골라서 진행할 것.

이건 작년에 했던 exercise와 같아. 작년에도 여러가지 프로젝트 중에서 하나를 골라서 커밋하고 진행했어 (blog post).

그것과 같이 우리는 지금 있는 데이터를 기반 해 한 가지의 most achievable product를 골라서 그 것을 만들 것.

작년 이 맘때 쯤 SpriteDX를 정했을 때 가장 좋은 점은, Goal이 너무나 명확했어. 뭘 하고싶은지 딱 머리속에 있었고.

몇개월 전에 이 exercise를 한 적이 있엇는데, 그때의 목표는 이랬어.

내가 unique하게 가장 잘 할 수 있는 cellular automata를 사용한 biome 제너레이션.

그리고 그를 통한 에이전트가 살아가고 상화 작용 할 수 있는 시뮬레이션 environment. 이게 컴페니 thesis야.

Future living AI agents need a place of belonging and place where their identity is formed.

이 company spiritus가 하나의 anchor고 또 하나의 anchor는 이거야.

Everyone works on agents. Memory system is getting better day by day. Near-conscious agents are evolving at lightening speed. There is no lack of talents in that space.

즉 sprited는 digital being 자체의 개발보다 AI-ready semantic virtual environment 만드는 것이 더 의미 있는 일이야.

If everyone is making agents, but they have no surface to land their agents on.

그래서 에이전트의 생물화(?)는 우리가 하지 않아도 여러 디시플린에서 일어나고 있는 현상이고 그 inevitability를 준비하는 회사가 되어야 해.

Virtual Environment vs Digital Being

There has always been a conflict between whether I should work on (a) virtual environment for ai beings or (b) ai beings itself (similar concept to ai companionship) or (c) something else entirely.

Building digital being is like making a ai vtuber IMO. For example, we could make a more advanced version of Neurosama and gain audience. Go-to-market would be near instant in this day and age and progressively we can build it out as I build audience. However, this was true for any other people. There was no significant strategic advantage for Sprited. At this time, everyone is building agents. Popularity is not necessarily a bad thing. The strategic advantage I have personally is the obsession over user experience and notion of privacy and security.

However, the model degrades very easily to what I call "AI-begger." The goal of the existence becomes a engine that extracts money out of users, and in return it gives people emotional values.

This market is easily most attractive market for indie developers because how easy it is to create a instance and build "their own memory schema." While writing this out, I still want to build my own AI friend that I can talk to. This is dream come true of most of us, and a game where many play and only few wins.

And the winning one won't be the most creative or even the most robust. Winning one will be the one that creates true connection with users and have a strong reach.

Given this setup, shall I still experiment further with digital being?

What we have so far

We have following demos.

A. Machi - Semantic Pixel World

Machi world system is basic biome engine for pixel world.

B. Anima34 - White Room

Digital Being-centric website where you can interact with a digital being.

C. Lunamachi - Cozy Room

Use SpriteDX characters for cozy room with music.

My take

One thing for sure is that I am not able to work on these projects in parallel. So I need to choose a project that sets up Sprited for success near term.

In deciding what to do, it is almost the case that I need to prioritize the company's stability rather than my desires.

What if I were to post all these projects on Reddit and do a vote. Ask them what people wants to see more from Sprited?

Comments

[Jin Sprited Sprited]

In a way, clarity supersedes creativity. There are so many different AI advancements happening all at once. Instead of dabbling in all of them it is important to find a niche and stick with it.