#social-game #conversation #humans #ai #idea
# [[Epistemic status]]
#shower-thought #to-digest
Last modified date: 2022-12-28 08:11
Commit: 1
# Related
- [[Social space AGI]]
- [[Conversation]]
- [[Conversations augmented by text-to-image AI]]
- [[New mediums of human interaction]]
# TODO
> [!TODO] TODO
# Langa 😛
https://langa.me
https://help.langa.me
https://discord.me/langame
https://cards.langa.me
https://chat-example.langa.me
%%
If you like Langa pitch, https://calendly.com/louis030195/rand
LessWrong: "The only path to [[Artificial general intelligence|AGI]] is reuniting humanity in cooperative conversations"
Investor: "Belonging is a problem that have no solutions since the dawn of time, why now? [[Artificial general intelligence|AGI]]"
Investor2: "Struggle to belong? We use the most powerful AIs to helps humanity have meaningful connections"
BasicHuman: "We helps you having more meaningful conversations. We show you active feedback & suggestions during your conversations" + add some buzzword
%%
>The best Tribal meetings include rituals that go beyond the cut-and-dried business of the company and help the employees get to know one another better as people, not just workers
>~ [[Reid Hoffman - Reid Hoffman - Blitzscaling - The Lightning-Fast Path to Building Massively Valuable Companies - libgen.li|Reid Hoffman]]
%%
# North star
## Customer value proposition
Langame: "We help invididuals and groups having more efficient and meaningful conversations by assisting conversations using artificial intelligence, reducing friction and augmenting human life time by 100x"
## Future vision
Langame: "Make every conversations more meaningful."
Langame: "Become America's leading human conversations AI assistant service with 2 billion daily active end-customers by 2030"
## Why you
Louis has always had trouble having interesting conversations, probably like other people we call introverts.
The first time he played conversation and language games, he really loved it, because it leads to very deep and open conversations leaving no room for superficial and unimportant topics.
Louis remembers going to an event in Paris, we had to bring pieces of paper with interesting questions written on them, and choose randomly, he loved it.
That's when Louis had this epiphany, technology was able to enhance conversations.
From there, he quit his job, built his idea while exploring other start-ups on and off. Over time, he piled up data and his AI improved. Despite all of these adventures, he still found himself on the same topic that kept popping into his head: the problem of conversations.
## TAM SAM SOM
TAM: 8 billion humans
SAM: 8 billion - access to a mobile computer with high performance internet
SOM: SAM -
## Unit economics
Unit economics (= profitability on a per-unit basis), e. g.: - LTV / CAC > 3 - >50% contribution margin - Create reasonable CAC at scale
## Timing
The existence of an essential, explainable and tangible change, current, or advancement creating tailwinds for your idea (e. g., of technological, socio-economic, or political nature).
- [[Software 3.0]]
## Competition
## Business model
## GTM Channel
## Growth
## [[Product-market-fit]]
## Fundraising
## Ask
My goal is to raise €____ in exchange for ____% of my company from an investor/investors who are ____. This money will allow me to hit milestones ____, ____, and ____ within ____ months and allow me to 1) raise my next round of capital and/or 2) achieve profitability.
# Strategies
How to achieve human-to-human AI optimized [[Conversation|conversation]]s?
## S1
This is one possible strategy I see:
### B2B (<5 years)
- personalised conversation starters (text, image, etc) for dating apps & communities
- personalised conversation starters optimised to generate the longest conversations between the two participants (i.e. dataset of (conversation starter, personality) <-> conversation (length))
### B2C (>=5 years)
%%
%%
#work-in-progress
# mess
https://langa.me
[[Langame 😛 on the metaverse]]
Invite me on **Langame 😛**, tag "lou", "loulou", "epi", ... (no schizophrenia I swear). Please guess I topic I like given this second brain ;)
In many-to-many [[Conversation|conversation]]s, any participant can influence the conversation direction, say, by talking about the weather or whatever.
The principle of Langame is to have a participant that is not human that influences the conversation in the best direction for all participants (excluding the Langame AI), given the understanding of every participant [[Mind|mind]], [[Personal growth/Goal|goal]]s, obviously.
Langame prevent [[We are delegating our intelligence to the machine - Artificial intelligence is eating organic intelligence]]
https://help.langa.me/artificial+intelligence+alignment
https://www.youtube.com/watch?v=G85GFQDWR_o
<iframe width="560" height="315" src="https://www.youtube.com/embed/G85GFQDWR_o" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
%%
<!--
Open source thoughts about https://langa.me
![[conversations.png]]
# Language ![[Language]]
# Conversation ![[Conversation]]
# Langame
Get access to the beta now at [langa.me](https://langa.me), be aware that the app is currently more suited to be used with known friends, i.e. people discovery is not properly implemented (recommendations, matching...).
Feel free to invite [[louis030195]] (tag loulou) between 8 AM UTC and 8 PM UTC, all my devices are disabled outside these ranges (and often within this range, when deep thinking, but with some luck you can reach me).
Langame is an attempt to solve and augment human beings in conversations using computers.
It is obviously not easy technically, but some things can be attempted.
![[Pasted image 20210604160433.png]]
### Evolution of the species
#### 06-05-2021
<iframe width="560" height="315" src="https://www.youtube.com/embed/a2hD0XsK5iA" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
#### 11-06-2021
<iframe width="560" height="315" src="https://www.youtube.com/embed/URLHBGbsAeI" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
#### 28-06-2021
<iframe width="560" height="315" src="https://www.youtube.com/embed/wmwRHWftAj8" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
#### 01-07-2021
https://productific.com/@Langame
#### 14-07-2021
I'm curious to build a simple, mini-[[AI psychologist]] within **Langame** to build a personality model of the user, necessary for the user match/reco feature.
Could do quick on-boarding, ask few generated questions. Should be pretty quick to implement
#### 18-07-2021
Would I use **Langame** in its current state? No.
I mean I already use it with some of my friends, every day, it's been great, but social features are too weak now.
I doubt humans can live without social interactions, it is imprinted in our [[Genome]]. I am thinking about social features, such as [[Groupism]], maybe scheduled **Langames**. I want to avoid large groups though, as I strongly believe [[Conversation]]s do not scale.
I like the idea of beta.spiel.com "achieve [[Fractal]] locality for communities".
I really want to play with the [[Brain]]-extending features, but I really need to work on social side.
#### 24-07-2021
I am extremely busy, but hopefully a browser extension for **Langame** is coming soon.
![[Drawing 2021-07-24 09.38.49.svg]]
Ah, and also **Langame** has been approved for OpenAI production 🎉.
#### 26-07-2021
early browser extension
![[Screenshot 2021-07-26 at 16.13.12.png]]
#### 01-08-2021
browser extension
![[Screenshot 2021-08-01 at 14.29.07.png]]
![[Screenshot 2021-08-01 at 14.28.54.png]]
![[Screenshot 2021-08-01 at 14.28.46.png]]
![[Screenshot 2021-08-01 at 14.28.26.png]]
#### 17-08-2021
Actually, I love **Langame**, but I must transmit this love to others :), **RECRUIT YOUR FUCKING USERS** a la [[Paul Graham]].
On another thought, I thought about a business model, because getting money would help grow the project:
- People can buy more tokens (to get **Langames**, i.e. conversation starters & more), it would be clearly disclaimed that when they buy tokens, it uses all the money to instantly generate some new **Langames**
- It would be possible to have a subscription also to make things automatic, and ofc would use all the money to generate more **Langames**
Another thought, I am considering the possibility of open sourcing Langame, as the goal of this project is for humans, not for making money, maybe open source could grow the project better than private + more money/business? Really early idea though, not sure doable yet.
>Without trade, innovation just does not happen. Exchange is to technology as sex is to evolution. It stimulates novelty. ~ [[Biology/Matt Ridley]]
#### 08-09-2021
Wow such delay. I have improved a few things, now **Langame** publish some thoughts automatically on https://twitter.com/langame_ai.
I am very busy with another project of [[Artificial intelligence|AI]], but hopefully can allocate some night and Sundays to:
- second-brain related features
- recruit users
- keep improving **Langame** [[Artificial intelligence|AI]] models
Especially as I see the "social app" thing something that I don't want and is not [[Business/Entrepreneurship/Innovation]], these works already well:
- https://gatheround.com/
- https://www.donut.com/
- https://lunchclub.com/
- https://www.quizbreaker.com/
[[Competition is for losers]].
##### Next
- **People can press to record themselves while conversing. The voice is turned to text. Later they can revisit their conversations, export it to obsidian.md, roamresearch.com or plain text**... I'd avoid recording other people imho?
- **Let people generate starters and edit content / topics. Then all are checked manually and confirmed/edited or not**. Later will build a filtering model, when enough data and time, I've done this for another project, fine-tuning OpenAI for this kind of classification. I pretty much **like lesswrong.com collaborative system of tagging, upvoting etc**. It gives me some thoughts, thought I'm still very concerned about how to scale generation without lowering quality. Say, Bob the random guy watching football all day might not enjoy the same kind of conversations than John the philosopher.
#### 09-09-2021
https://blog.readwise.io/readwise-reading-app/
soon:
- export conversations text to readwise
- export conversation starters to readwise
#### 10-09-2021
**exports to readwise basically works now**.
Though it could be more automated, i.e. connecting Langame account to readwise.
near future:
- page to let **people generate [[Memetic|memes]]** and improve content / tag, then Langame team would check & confirm it, the users get rewarded (**Langame coins**) for providing high quality questions.
- to scale up this, people could become **Langame** reviewers, plus of course several **fine tuned models for filtering quality**.
#### 12-09-2021
<iframe width="560" height="315" src="https://www.youtube.com/embed/6p3s7M5fLRE" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
#### 17-09-2021
- Actively recruiting users, lesswrong.com might be an interesting hive.
- New feature to record meme during Langames (speech to text), you then can digest them, export to readwise etc. ![[Pasted image 20210917100330.png|200x400]]
- I need some ideas on how to integrate readwise/obsidian/roamresearch more closely with **Langame**, please send ideas :p
[email protected]
- ![[Screenshot 2021-09-17 at 15.49.59.png]]
- Will balance that a bit :p
#### 18-09-2021
<iframe width="560" height="315" src="https://www.youtube.com/embed/1FGdzDMfLBg" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
see: [[Baldwin effect]]
#### 11-10-2021
- registration soon (US company, probably Stripe Atlas), 2/3/4 cofounders
- pivot to text only or text + audio (various reasons, it's easier to augment semi-asynchronous, non-real time conversations)
- web version works well, might be publicly released, might be interesting
- going back to the "augmented conversation" more than "second brain", we want to augment most people conversations, and eventually see if we can still manage to expand people knowledge
- I'll go to silicon valley sooner or later (networking etc.), probably intermittently
- soon co-founding https://www.linkedin.com/company/inkdrop-ai too. I am transparent with my co-founders, I strongly believe I can do both, I'm quite good at time management, focusing and [[Deep work]]
# Ideas
- let people throw an idea, a quote (readwise.io), something into a box, it create a convenient starter based on it -> **MORE COLLABORATIVE FEATURES**
- adjust starters/topics/app to **people personality**
- exporting audio also? (instead of speech to text)
- connect plenty of apps (**twitter share Memes, Readwise, Obsidian, Notion, Roam, Youtube?**) tighter than the native share thing which adds friction
- With the slow transition from big physical computers, human-machine interfaces, to [[Brain Computer Interfaces]], currently, earbuds (Apple launched it as always), we could give **conversation feedbacks into people ear** as audio ("you should be nicer", "try to ask this, that...")
-->
%%