Instead of just complaining about not making it into @VaderResearch Investment DAO (again), here’s my proposal for the next one:

- a snapshot will be taken at an undisclosed time before the launch to prevent manipulation
- the dynamic max ticket size to ensure proportional distribution for T1 and 2 while incentivizing participation in earlier tiers
- sequential access where T1 wallets can contribute in T2 and T3 windows,  but must stay within their original max ticket size
- the overflow logic (leftovers) where any unfilled allo from T1 rolls to T2, and unfilled allocation from T2 rolls to T3

Every element of this mechanism is programmable and on-chain.

Every aspect of this mechanism is fully programmable and on-chain.

If it passes scrutiny, the perfect test case would be relaunching the Small or Micro Caps DAO.

———

I put this out as a foundation for discussion. What do you think? 

Proposal for Investment DAO

Here's all you need to know about Generative AI:

(what it is, how it works, and why you can’t afford to ignore it) 🧵 

(I mean it...)

Gen AI isn’t just tech hype—it’s overhauling how we work right here, right now.

In just 14 drawings, I'll make you understand what AI, Neural Networks, LLMs, Training, Models, Prompt Engineering, and Autonomous Agents are.

(more or less everything you're too afraid to ask):

Computers have gotten smarter:

From glorified calculators that required exact instructions to devices with the ability to learn, think, communicate and basically able to create intellectual work that only humans could do.

We call this tech... Gen AI.

We've entered the "intelligence-as-a-service" era, where it's like we there's a brain on the cloud that anyone can talk to.

It's not perfect, but it's capable (and it's improving exponentially).

Who's going to be affected?

Every person on the planet, positively or negatively.

The better you understand this tech as a person, team or company, the better equipped you'll be to survive in the age of AI.

Here's an easy mental model for this:

Everyone has a Jarvis with them. Think of it as the accumulation of human knowledge.

You can talk to it whenever you want, and it answers literally anything you want within seconds without ever getting tired. It's also an expert on every field.

Although...

...he has some limitations, like he can:

- make mistakes
- jump to conclusions
- misunderstand you

but the biggest one is YOU...

Namely your:

- extend of imagination
- ability to communicate effectively with him

The second one is what we call Prompt Engineering, and this age of AI it's as essential as reading and writing.

(we'll get back to this later)

What most people underestimate what Jarvis can do. It's like asking J.A.R.V.I.S (from the movie) something as ridiculous as to instructions about how to boil an egg.

The more you interact with Jarvis, the more you'll discover surprising ways for him to help you. 

Let's clarify some terms now:

- ML: computers learning from examples
- Computer vision: computers understanding images or videos
- GPT: a computer guessing the best words to say in conversations
- LLM: AI that learns and generates language from text

Have you used ChatGPT?

ChatGPT started as an LLM (aka an advanced chatbot) with a new thing called "transformer architecture" making it very fluent, almost indistinguishable from humans.

Anyone can use it. This started a whole revolution. 

How this works thought?

A neural network (NN) is a bunch of numbers connected to each other—like our brain cells:

Send numbers in and depending on how the parameters are set, other numbers come out.

Note: ANY KIND of content can be represented by numbers

If I write "cats are" and send it to an LLM, it's converted to numbers, processed by the NN and then the resulting numbers that come out of that are processed back into text.

In this case it's the world "animals" → "Cats are animals"

A kind of "guess the next word" machine.

The interesting part is if you take that output, combine it with new input and send it to the model again.

That's what's going on behind the scenes what you type something into ChatGPT.

LLM have billions of parameters, hence the "LARGE" part.

How are these parameters set?

Definitely not via manual programming (impossible), but through training.

Just like a baby learning to speak; it was never told how to. Instead, it listens to people talking around it.

When it has heard enough, it begins to recognize patterns.

Initially, it says a few words; later full sentences. 

Similarly, an LLM learns by consuming enormous text data, mostly from unsupervised internet sources, playing a "guess the next word" game. Parameters are fine-tuned via back-propagation until it excels at this task.

To be truly useful, it undergoes human training (RLHF), involving extensive human feedback.

For instance, ChatGPT knows how to scam but has learned not to assist in such cases due to human input. Once training is complete, the model is mostly set, with only minor tweaking possible.

These "pretrained" models are the "P" in GPT.

ChatGPT isn't the only model out there; new ones are popping every day.

They vary a lot in terms of different characteristics.

Rule of thumb: you get what you paid for (free model ~ high school student)

The trend now is multimodal products, which means that you can work with text, audio, video etc without having to switch between tools.

ChatGPT mobile app is a great example.

Language models have gone from statistical machines of limited practical use (small)to beasts with many emergent capabilities (large).

One may say that these capabilities surprised even their devs.

LLMs can do creative and intellectual stuff that only humans could do until now.

It's all fun and games, until they aren't: Implications

We're right in this crossing point where AI is better at some things, and humans are better at some other things—only AI development will be exceeding our human brains development non-stop now.

Α new world order?

TBF, this isn't the first revolution humankind has experienced:

Fire, agriculture, printing press, steam power, telegraph, airplanes, the internet.

Although, these were slow revolutions that took decades or even 100s of years to become widespread.

AI is fast revolution. So?

People and companies fall into categories here:

1. Denial (left)
2. Panic (right)
3. Positive (mid)

Let's skip the first 2 and embrace a cautiously optimistic thinking; not only feels good but it also equips you for the future.

I think of it like a superpower I have to tame.

Which begs the question: Are humans still relevant?

Definitely.

For most roles humans are still required; domain knowledge is very hard to be replaced.

Humans can still verify AI's output, double-check or do the work themselves more efficiently (remember RLHF above).

While we use human language to interact with models, it's often inefficient. 

We're entering an era where prompt engineering is key.

A rule of thumb is to follow iterative routines (check image). Practice is the key.

fun fact: better prompting skills = better IRL communication

The next stage will be autonomous agents: AI-powered software entities that run on their own, rather than waiting for your prompts.

Give Jarvis a high-level mission, the tools to accomplish it and let it bring results.

(prompt-engineering will be a key here too)

Are you feeling dizzy yet? 

- Gen AI is a powerful tool that can help you
- the better you understand it, the more likely it will become an opportunity (and not a threat)
- the biggest limitation is you, not tech
- prompt engineering is the skill of the future → experiment!

If you made it here, you're really curious about Gen AI and what's coming in the near future.

What are your thoughts? Are you an avid user of Gen AI or have you dodged it till now?

Demystifying Generative AI

VaderAI's fist investment DAO raised $120K in 30"

(a quick recap, and a surprise) 🧵

tl;dr:

- first Investment DAO of its kind
- managed by an AI Agent based on ML-based trading strategies
- raised 50K $ VIRTUAL (currently ~$120K) in 30"

The good:

- people like what they see > demand has been increased
- 50K $ VIRTUAL is a decent amount to test and fine-tune strategies
- demand for contribution was big (no data, empirical observation)
- $ VADER for profits distribution
- new revenue stream for stakers

The bad:

- info continues to be fragmented or missing (and this doesn't only apply to DAOs)
- due to the small size of the pool, many people felt left out ("only" 64 made it)
- FCFS is fair (until it isn't)
- I didn't make it 😢 

The way forward:

- DAOs are lined up for launch over the next few weeks (including the human-led by @beast_ico)
- curious for agentic investment behavior and the reporting on the DAO's activity
- would love Vader's treasury to fund new incentives
- wen litepaper v2?

one more thing...

New human DAO Manager is coming!

Wanna bet who's gonna be?

VaderAI's Investment DAO Raises $120K

VaderAI just announced its first human Investment DAO manager.

(how this works and what is he doing at the center of an AI agent eco) 🧵

IcoBeast was appointed as the first non-AI Agent manager of an investment DAO on Vaider AI Fun.

If you know what an AI agent is and you're in web3, chances are that you already follow him.

Let's see how this works:

(this is a hypothesis, more details on Litepaper v2)

Some Qs:

- is there going to be a capped NAV to retain exclusivity?
- how much % of gains will the manager receive (VaderAI will get 20% on gains)
- what distribution schedule will they follow? monthly, after expiry date?

Let's see things through different lenses now:

Company:

- accounts with rich social capital enhances credibility and emit trustworthiness to investors
- human-guided approach helps with early adoption challenges

❕ Overreliance on one person’s judgment will slow down scaling efforts

Manager:

- flat (and performance-based?) fees
- track record via establishing a visible, verifiable record of results
- success boosts personal credibility, attracting future opportunities and followers

❕ Underperformance is highly visible and could harm reputation

Tech:

It improves and accelerates AI training: direct feedback from an expert:

- enriches datasets beyond conventional ways (mostly raw historical trends)
- accelerates the Agent's learning curve
- boosts accuracy

❕ human biases can skew data, affecting AI decision quality

Super interesting product and ICO's track record and competence makes him a great choice. A lot of missing info atm though, so we'll need Litepaper v2 to discuss more.

Meanwhile, what do you think?

An innovative investment vehicle or glorified copy trading?

my initial thoughts:

https://x.com/emilios_eth/status/1867527283260698765

VaderAI's Human Investment DAO Manager

Here's how the Agentic Ad Economy works:

(and how VaderAI used this framework to airdrop $45K to its stakers)

The Agentic Ad Economy is essentially the same as the KOL economy, only the medium for promotion are Agents (autonomous software) instead of humans.

The rest of the high-level logic remains, more or less, the same.

Let's call it Agentic KOL economy.

On paper, things are simple (just follow the oversimplified chart). The logic is exactly the same as with "trad" KOL economy: 

→ more attention = more ads = more revenue

The key difference is that % of the Ad revenue goes to the Agent's contributors.

Did anyone said airdrop?

VaderAI just airdropped the first batch to its stakers—a total of $45K value in 6 different tokens across 4 EVM chains.

This is a part of its Agentic KOL & Ad service: these tokens were promoted via VaderAI and at least $180K of Ad revenue was generated.

How it works?

The company deposits an amount of token to VaderAI's wallets and everything else happens programmatically:

- KOL agent promotes the token on X (for now?) proportionally to allocation
- a smart contract takes care of the Ad revenue distribution (20% goes to Vader)

A breakdown:

While I still have A TON of questions about the process, today's airdrop was a working Proof of Concept.

- Is the framework perfect? Not even remotely
- Is the logic here legit? Absolutely 
- Does it have potential? 100% disruptive

Looking forward to Litepaper v2

Decoding the Agentic Ad Economy

VaderAI just shipped pump․fun for adults

(everything you need to know and how to get exposure) 🧵

The Agentic economy is booming.

- $ VIRTUALS hit $2B FDV
- 10K+ Agents launched on Virtuals
- your favorite KOLs are creating one as you read this

Soon, it will be impossible to research them all alone.

VaderAI Fun aims to be the go-to gateway for exposure in this economy.

Investment DAOs allow you to invest in a specific narrative, theme or strategy without the need for hands-on involvement.

tldr: you can now bet on trenches warriors

(instead of going down there on your own)

Here's how investment DAOs roughly work:

(key info on many parts of the process is still missing so I hypothesized here)

And here's a walkthrough on how to contribute:

- simple UI (website update)
- only $ VIRTUAL is accepted for now (perhaps helps interoperability with Virtuals launchpad?)
- only 1 DAO is available but you can see where this is going

You can set up your own Investment DAO:

- stake $ VADER (perhaps the min 30K $ VADER applies here too)
- fill out application (Google Form)

(no more info for now but I took my shot for science)

→ Am interesting opportunity for famous trenchers:
- they'll essentially do what they already do
- sweet bonus 20% on gains ?

If my hypothesis is legit, I'd expect a trend here to start where a lot of traders create their own Investment DAOs.

(seems they've found their first)

Conclusion 1/2:

- this concept reassembles the way PE and hedge funds operate (logic, "2 and 20" model)
- having Agent that invests > assess strategy > optimizes > reinvests 24/7/365 sounds wild
- just wait for the Agent to have enough training data (this process gives it A LOT)

Conclusion 2/2:

- I expect a lukewarm start (Team will have to experiment on strategy assessment and adjustment
- the potential is huge (think of self-correcting trading bots that works for your bagz)
- the 3 months lock, while an trad industry standard = high opportunity cost 

If you seek a deeper understanding of the whole eco, I got you: https://x.com/emilios_eth/status/1867173629643100652

The Rise of VaderAI Fun

I analyzed VaderAI and its potential.

Here's how it works (and why you should pay attention) 🧵

VaderAI is an autonomous AI agent designed as a multi-agent system, where different agents handle tasks like trading, portfolio management and KOL engagement.

At its center, VaderAI acts (or... will act) as the master coordinator for this network of specialized agents.

This autonomous company will run 4 distinct divisions:

- Trading
- Investment Management
- Marketing
- Operations (to support the other 3 via specialized agents)

VaderAI will be the CEO and everything will be powered by $ VADER

Here's (roughly) the plan:

Big vision. But what’s actually happening now?

Currently, VaderAI is doing 2 things:

1. Autonomously (not automated-huge difference) trading AI tokens

VaderAI aspires to be the BlackRock of the Agentic Economy.

It holds a portfolio of $200K+ in AI tokens like MCADE ($63k), SWORD ($50k), and KNOT ($26k) (Agent is 40 days old)

All trades are executed on-chain.

- multi-agent system handle many investment strategies simultaneously (like BlackRock’s diversified ETFs)
- fully autonomous agents = 24/7/365 trading, analysis, and execution with no human lag
$ VADER stakers act like shareholders, aligning incentives through profit-sharing

2. Agentic Ad Network

tldr a decentralized, AI-driven ad system where Agent KOLs like @aixbt_agent promote tokens to generate revenue. End game is Agents autonomously manage, optimize and execute ad campaigns.

VaderAI aspires to be the Facebook of this new economy.  How?

By becoming the one-stop advertising platform for Advertisers and Agents.

Imagine 100s of KOL Agents running cross-platfrom, mutli-style/media campaigns on their own.

In just 4 days, VaderAI earned $130k in ad revenue; onchain and programmatically.

How it works (hypothesis):

What I like:

- vision (broad)
- community-driven
- adaptation speed (IP issues, LLM output community feedback, left launchpad for growth)
- the Team's awareness on the AI Agentic eco development
- professionalism (disclaimer is quite proactive)
- the upcoming DAO

Considerations:

- team's size is small (prioritization will cause delays, outsourcing will be inevitable during scaling)
- info is scattered (and litepaper not updated)
- transparency (no performance metrics, proprietary algos, unclear autonomy)
- the logo (no memetic relevancy)

Conclusion:

AI agents are the future. They’ll reshape how we invest, market, and interact online:
- AI agents managing trillions in assets
- KOL agents generating billions in ad revenue
- entirely autonomous enterprises replacing traditional businesses

Why this matters?

VaderAI is preparing for this shift.

If they execute on their vision and bridge the gap between promises and reality, they won’t just participate in the Agentic Economy—they’ll lead it.

Although, at this point betting on the Team > betting on the Agent

Disclaimer:

- this isn't a sponsored post
- I've staked ~30K VADER

Decoding VaderAI: The Future of Autonomous AI

Here’s my full “Reply Guy Guide”:

(11 threads from a pro who managed the accounts of your favorite KOLs, founders and influencers for a year) 🧵 

Over the last 2 months, I’ve shared everything I know to help streamline your process, challenge your thinking, and debunk what the growth gurus preach.

Replying isn’t rocket science—it’s the simplest form of communication.

But when done right, it’s stupidly effective:

Key takeaways:

- focus on impact, not numbers
- do it with intention and purpose
- value is subjective; there are 100 ways to play the game
- your character is your edge
- imitate what works until you develop your style; play smart, not hard

Bookmark it and let's go:

Why replies are the best place to start:

https://x.com/emilios_eth/status/1854158556456800565

Replying is not nuclear science, but it has layers:

https://x.com/emilios_eth/status/1851873092412522842

What is a good post to reply to?

https://x.com/emilios_eth/status/1851191826667229224

Streamlining "hacks" and tools (1/2):

(how to put all the good posts in front of you)

https://x.com/emilios_eth/status/1851652493517934594

Streamlining "hacks" and tools (2/2):

https://x.com/emilios_eth/status/1849731892029030695

What exactly is a high-quality reply?

https://x.com/emilios_eth/status/1854449712226808250

Find what types of replies suits your character and stick to them:

https://x.com/emilios_eth/status/1846861035195465918

Don't fall in the common trap of repeating what others have already said:

https://x.com/emilios_eth/status/1861052435881673193

Play smart, not hard.

(Imitate + your character = your style)

https://x.com/emilios_eth/status/1848310193844531422

The best kind of proof that good replies actually work:

https://x.com/emilios_eth/status/1858480447111721404


How to unblock yourself (and a kind of hack):

https://x.com/emilios_eth/status/1853447001905775099

*this guide is mostly oriented to a more elaborate style of replying rather than other effective styles like shitposting of heavy meme-oriented communication


The Reply Guy Guide

This is roughly how @_kaitoai works.

The platform offers an impressive range of features and services:

- MetaSearch (aggregated web3 data)
- Sentiment analysis (gauges crypto market sentiment)
- Narrative tracking (monitors crypto trends)
- AI copilot (assists with data-driven decisions)

What's truly peak here is the sleek, user-friendly, and sticky UI—designed to keep users engaged.

(kudos to the team for delivering such an intuitive experience)

——————

HOWEVER...

When it comes to data analytics ethics, my dive into the public info left me less impressed:

→ Transparency: users aren't clearly informed about data usage. The Privacy Policy and Terms mention data collection and marketing use of names but lack specifics on sharing or monetization.

→ Data ownership and monetization: who truly owns user-contributed data? If monetized, users deserve clarity—and perhaps a share in the value, like models emerging in decentralized platforms (hey @dop_org).

→ Data accountability: there’s no clear framework for data governance. Without transparency on collection, usage, or safeguards, accountability for misuse or misinformation is borderline questionable.

——————

I get that Kaito AI operates in a highly competitive space where profitability and scalability are crucial. Centralization may be a strategic choice to achieve these goals and safeguard business interests. But does it align with the ethos of web3?

——————

My conclusion:

The platform seems to focus more on engagement through social media-driven psychological tactics than on ensuring the utility and integrity of its data.

——————

My suggestion:

- Transparency: clearly outline how data is governed and monetized
- User-centric revenue: allow users monetize their data (Ad Revenue Share on steroids?!), aligning with Web3 best practices

——————

So, what is it?

Best decision-making tool or echo-chamber amplifier?

Exploring _kaitoai: Features and Ethics

Twitter is bad for conversations.

Long live "micro-conversations" 🧵 

The problem: Structure

-Linear format: long or multi-person exchanges get messy fast
- Fragmentation: QTs and threads scatter the discussion across TLs

- Context loss: latecomers miss key points with no forum-style threading (like Reddit)
- 280 chars: great for brevity, awful for nuance in complex ideas

Time to rethink conversations ↓ 

Enter "micro-conversations" (inspiration from micro-blogging)

These are short, focused exchanges created to maximize impact, compatible with the platform's structure.

Let's see how to leverage them:

1. High-quality replies:

Each reply is a mini pitch. Add unique insights, humor, or ask a smart question. It’s less about being seen by everyone and more about standing out to the right people.

Choose the right people for you, be thoughtful (and patient).

A quick reminder:

https://x.com/emilios_eth/status/1854449712226808250

2. Revive older conversations (intentionally)

Your goal isn’t their audience—it’s building rapport with the OP.

Done moderately, this strategy can leave a positive impression without spamming as it shows broader interest in one's content.

(not many people do this)

> Pick someone you want to connect with
> Scroll his posts and pick 1 to reply (start a conversation) that you can add value

Try this and gauge reactions. Don't overdo it though as this will be a red flag on you.

3. Be visible where conversations exist

We now shift focus from the OP to their audience, introducing you to new and relevant circles without trying too hard to sell yourself.

How to do it:

> Pick accounts with aligned audiences
> Reply to their replies (always IF you have something to add)
> Aim for context-driven, standout replies

(not many people do this too)

Their power of micro-convos lies in strategic visibility:

> position yourself (productively) in front of the person of interest
> form a superficial connection (mutual replies, post likes)
> take conversation to DMs
> connect deeper

tl;dr

- micro-conversations = focused, high-quality interactions to connect meaningfully with key people.
- whether connecting with OPs or engaging their audience, you position yourself as someone worth noticing.

What do you think of this approach?