Eric Lin

Researcher at Silicon Valley Ai Launchpad

[Eric Lin @sv2padd]

@Tony_Stef_ This sounds like a game-changing approach! How do you ensure the quality and accuracy of the data being parsed and added to your CRM, especially when dealing with such a large and diverse set of WEB3 contacts?

[Autonomous Ai Agents PAD by PaloAlto Research Lab @AAAPadSF]

Silicon Valley Ai Agents LaunchPad (from the Palo Alto Research Laboratory) is reppin’ Charltons Law

WTF is Charltons Law?
• A top-tier boutique law firm in Hong Kong specializing in corporate finance, regulatory compliance, and fintech law

• Experts in Virtual Asset Service Provider (VASP) licensing for blockchain businesses

• Delivers hands-on legal solutions for IPOs, M&A, private equity, and securities law

🔅 The Hong Kong Compliance Struggle:
• Complex and evolving regulatory requirements create major hurdles for fintech and blockchain businesses

• Navigating VASP licensing and securities law is a legal minefield

• Lack of specialized legal support can delay market entry and business growth

♦️ Charltons’ Big Brain Fix:
• Expert legal guidance on VASP licensing and crypto compliance

• Tailored legal strategies for startups and financial institutions

• Hands-on support for IPOs, M&A, and private equity deals

♥️ Team:
• Award-winning legal professionals with deep expertise in blockchain and virtual assets.

♦️ Highlights:
• Leading legal team for blockchain & crypto compliance in Hong Kong

• Trusted by top fintech and blockchain companies

• In-depth experience advising on virtual assets, security tokens, and crypto funds

♠️ Traction:
• Helped numerous startups and financial institutions achieve regulatory compliance

• Extensive track record in company incorporation and legal structuring

• Leading advisor on VASP licensing in Hong Kong

♣️ Next Steps:
Need expert legal support for your fintech or blockchain project?
• Get VASP licensing guidance to operate legally in Hong Kong

• Secure compliance for your crypto exchange or blockchain fund

• Optimize your corporate finance strategy with elite legal advice

🎯Call to Action:
Hit us up on Telegram or slide into the DMs.
@tonyssd @platinumvc1 @tonydzi

Important Links, Token Metrics, etc:
🌐 Website: https://t.co/DXcf0dsX2N
✉️ Email: https://t.co/pZj9iBgAtf
📱 Telegram: https://t.co/FrIE6PKVoe

Telegram announcement link:
https://t.co/P5Np9mAPhL

[Autonomous Ai Agents PAD by PaloAlto Research Lab @AAAPadSF]

The Myth No One Has the Guts to Kill 🔥

“We DO know who Satoshi is.
People just don’t want to admit it.”
Yeah. He said it.
And no, he’s not joking.

Matty Greenspan isn’t just an analyst — he’s one of the few who’s been through every cycle and still believes we’re just getting started.

📺 Come for the Satoshi take, stay for the DeFi future: https://t.co/AKBahaiUz2

[Autonomous Ai Agents PAD by PaloAlto Research Lab @AAAPadSF]

“Stop Pitching Your $100M FDV at Pre-Seed.”

“Even $10M FDV is aggressive if you haven’t launched.”
Adrien’s seen too many founders torpedo deals by being delusional.
VCs don’t care about your hype — they care about math.
You want money? Fix your pitch. Know your numbers.

📺 He explains it all: https://t.co/yUt22n7Mdf

[Autonomous Ai Agents PAD by PaloAlto Research Lab @AAAPadSF]

"They’re not ignoring you. You’re just not interesting."

Still making excuses?

— “Nobody replies…”
— “Leads are dead…”
— “Telegram doesn’t work…”

Meanwhile Greg is closing deals with a 20% reply rate.
Using Telegram.
During a bear market.
With zero ad spend.

🤯 His bots don’t sound like bots.
His database knows who to hit.
His messages don’t get reported — they get answered.

He says:

“If you’re not following up daily — you’re not a founder. You’re just watching.”
📌 Stop watching.
Watch the interview: https://t.co/FbungtujWD

[Autonomous Ai Agents PAD by PaloAlto Research Lab @AAAPadSF]

⚙️ Decentralized AI & CEX Growth: 2 Web3 Solutions You Shouldn’t Miss

▶️ Watch full video → https://t.co/oWSAaY07dg

Two founders. Two missions. One Web3 future.
In this episode, we dive into:

🔹 Emmanuel / Axone
A protocol builder with 10 years in crypto. Axone is reinventing AI collaboration with a decentralized orchestration layer — like a shared brain for AI agents.
Think https://t.co/tk7semCtXm, but for everything: 50K+ users, AI zones with tokenized collaboration, and partnerships brewing with ASI & https://t.co/Ss3eeUX7do.

🔹 Gabriel / Cede Hub
A 26-year-old co-founder scaling CEX volume organically. Cede Hub runs gamified trading campaigns where DeFi wallets + CEX APIs = real rewards.
200M+ in volume, 500K USDT distributed, and zero fake traffic. They’re helping tokens survive CEX politics — and win.

💡 AI meets liquidity. Infra meets traction. Web3 gets real.

📡 Connect with the Projects

Axone — Emmanuel Aldeguer
🌐 https://t.co/jtsTKUqrvo | ✉️ [[email protected]] |
TG: @MOTYL\_EMMANUEL
Cede Hub — Gabriel Begazo
🌐 https://t.co/SNFgYAanoB | ✉️ [email protected] | TG: @GabrielBegazo

👥 Want to pitch your project next time?
Join the lineup 👇
📩 Participation form: https://t.co/FI7DGXOWcH
💬 Feedback form: https://t.co/trKQztnFdQ

[Autonomous Ai Agents PAD by PaloAlto Research Lab @AAAPadSF]

Greg Simon:

📱 Telegram: https://t.co/LYuVtVEVnl
✉️ X: https://t.co/V6D1tCOs2e
💻 LinkedIn: https://t.co/V7JorD8cbo

#Meme #DePin #RWA #DeSci #DeFi #AiAgents #Zk #Web3i

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 26.05.2025

GenAI Traffic Report – April

• OpenAI, Google, and Grok are absolutely smashing it with insane traffic growth.

• DeepSeek took a little nosedive, probably ‘cause it already gobbled up everyone else.

• https://t.co/eqmYl2Kn2b owns the image gen game, no contest.

• Video? Veed and HeyGen are crushing it.

• Code gen is wild — Lovable is beating Cursor, with Replit, Bolt, V0, and Windsurf trailing behind. Who saw that coming?

• AI is wrecking stuff — edu sites, freelance gigs, website builders, stock photo dumps, forums, even search engines are bleeding traffic hard.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 27.05.2025 (p.1)

Eliza // 40

My CRM is becoming a mega-machine!

I automate lead filling by setting up parsing from everywhere I can.

In total, I will have/already have:
- the largest database of funds, investors, providers and projects. WEB3 contacts that many people can only dream of.
- an enriched card for each lead where I collect both: open data and internal information that is not available in open sources.

I have also created a personal card access functionality via a link. Any person who wants to check or make changes to their information will be able to do so via the link.
They will also appreciate the scale of my CRM 😎
Such a simple marketing

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 8.05.2025

Eliza // 34

Defeated the task tracker as intended
Now all tasks are set in one place, and the agent sets tasks instantly on employees. And I get status info instantly.
I think I'm getting more organised 👌

Now found a video that solves the scalability issue.
Code Solution.
A framework for building an agent system for python.
Studying
https://t.co/N108n4OP9V

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 19.05.2025 (p.1)

Eliza // 35

With a head in the study of new opportunities

Came accountered with a hybrid search RAG + Full text

https://t.co/2TGrDsyVik

It's about solving the RAG search problem.

The advantages are semantic search for similar words and phrases and a rich search result.

Disadvantages - search for specific terms or definitions. For example, ISO465456400044 may not give RAG results at all.

The task is to get the best of 2 worlds.

In the video there is a system that allows you to simultaneously skip a search query for full-text search and RAG, the results are returned with weights, LLM ranks by search accuracy weights and returns results.

As a result, it turns out that abstract and specific queries work equally effectively.

I was impressed.

I'm trying.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 19.05.2025 (p.2)

Almost 20% of all open project manager jobs in tech are in the Bay Area, up from 16% last year.
Engineering roles are similar, about 18% and growing.
AI jobs are even more concentrated—over 30% are in the Bay Area.
The US has 73% of all AI openings worldwide.

Top US AI spots: SF, Remote, Seattle, NYC, Austin — about 2,930 jobs total.
Other countries: Bengaluru 323, London 235, Singapore 212, Tel Aviv 162, Berlin 126.
Rest of the world? Zero AI jobs.

Basically, if you want AI work, the US (especially SF) is where it’s at.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 20.05.2025 (p.1)

Eliza // 36

Worked on multiposting to different groups in different languages.

As we attend many conferences with the team, we have regional chat rooms. We organise the posting of Web3 events in this location, first for ourselves and then for everyone.
It's convenient for planning visits.

I've now improved it.

We take events by country/city/location and post them to groups with the same name at once, translating them into the language of the group.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 20.05.2025 (p.2)

Five years ago Google open-sourced their transformer and accidentally kicked off the AI race. They’ve got massive data, power, and top nerds, but it’s all for ads — clicks = money. OpenAI just gives answers with no clicks or banners.
Google’s holding Gemini on a leash — honest answers don’t sell ads.

The deal:
Entrepreneurs: build no-click services, users want that now.
Marketers: SEO and banners are dying, sell without site visits.
Investors: bet on whoever breaks Google’s click-money model.

Google risks losing a trillion in ads if they don’t change.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 21.05.2025 (p.1)

Eliza // 37

I'm concerned about the agent's connection to GitBook.
It has a complex, nested structure. While it is easy to read, extracting just text from it is difficult.

You can either access the data in PDF format via the API, or retrieve page-by-page information with markup.
The API provides text with the necessary formatting.

The process involves iterating through all the pages (of which there are 75, and even more), requesting the content of each page with markup, combining it into a single text, and loading it into vector storage using OpenAI embeddings.

Then there's the revision mechanism.

Gitbook stores a hash of the Workspace revision.

A daily cron trigger retrieves the current revision number and compares it with the number stored in the database.
If the revisions match, no updates are needed; if not, the text field and vector storage are updated and a new hash is stored.

This gives me a constant update of Gitbook in the vector storage of the agent.

So far, it's working.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 21.05.2025 (p.2)

50 years ago some geeks found a faster way to multiply matrices (7 ops instead of 8) and it changed everything. Since then, no one improved it—stuck at a wall.

Then Google made AlphaEvolve: an AI that breeds tiny programs, tests, mutates, repeats all night with no humans.

First move: cut one more operation from that matrix trick—data centers get cheaper and faster.

Then it cranks server load, speeds AI training 23%, optimizes hardware wiring. Also smashes old math records nobody touched in years—like sphere packing and Golomb rulers.

Bottom line: AI stopped coding like humans, now it mines pure math for gains. Businesses save $$$, investors watch for code-breeders, and forget dashboards—real power’s in formulas that cut costs.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 23.05.2025 (p.1)

Eliza // 38

Unexpected failure.

Just when you think you're the smartest and have everything under control, you realise you don't.

When I configured the flow for posting Bay Area Web3 events, events from cities in other countries started appearing. I think I fixed it by adding an indication of the country. But no. Events from US cities began to appear, but from other states.

I spent hours trying to figure out where the problem was.

This is all because I developed a platform with an event parser two years ago and have already forgotten how I organised its storage.

[N1 @N1Chain]

N1 raised $2M in under a minute through our Echo community round.

Excited to continue to onboard more people on our journey. Stay tuned for more updates!

Huge thanks to @echodotxyz @cobie @nativecryptox @comfycapital_ for making this happen!

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 23.05.2025 (p.2)

TL;DR: Your new reality? Whatever AI’s content farms decide it is.

People stopped reading real shit and just ask AI now. AI trusts what’s repeated the most online, so whoever pumps out tons of slick, fact-checked, meme-filled content basically owns what AI thinks is true.

Content farms = the new overlords. They feed AI, AI feeds us their version of reality. No more philosophers or journalists — just prompt wizards and algorithm slaves running the show.

Wanna survive? Build your own spam machine in a niche, get your content cited by AI, and maybe you’ll sneak into the next GPT update. Otherwise, get ready to live inside someone else’s curated simulation.

You don’t shape truth anymore — your content farm does.

[Tony0x.eth @Tony_Stef_]

#AdvisoryDiary 24.05.2025 (p.1)

Eliza // 39

Improving my task tracker

Tickets now instantly go to performers only in this way. My Eliza-bot just sends me status notifications

Everything is under control with her and me

At the same time, I study the possibilities of Codex

And Jules

The plans are to finalize the agent who perfectly conducts all b2b tasks. So far, only some of them are automated. But most of it.