Abnormal AI

A different kind of insider. Scanning Web3 for AI that break the norm.🔥

[Abnormal AI @AbnormalAIX]

Why Autonomous AI Agents Will Reshape Web3 Adoption by @AbnormalAIX

Web3 offers real ownership, transparent finance, and digital freedom. But adoption is still slow. Most people give up because of seed phrases, high gas fees, confusing interfaces, and the need to watch charts 24/7.

Autonomous AI Agents solve this.

These are smart programs that can act on your behalf on blockchain - trading, optimizing, and managing tasks automatically.

This could be the biggest onboarding wave yet. Let’s break it down logically.

[Abnormal AI @AbnormalAIX]

Design is becoming a coding problem.

The "Visual Code Generation" stack is forming fast:
- Browser-native → Claude Design, Cursor, Figma
- Vector-native → Figma, QuiverAI
- Motion → LottieGPT, OmniLottie
- Video → HeyGen HyperFrames, Remotion
- 3D/Simulation → BlenderMCP, VIGA

Designers who don't speak code will be replaced by code that speaks design.

[Abnormal AI @AbnormalAIX]

Microsoft quietly launched 7 MAI models at Build 2026:

- MAI-Image-2.5 - image gen + editing, #3 on https://t.co/jD13uMUf2W
- MAI-Image-2.5-Flash - faster, cheaper variant
- MAI-Transcribe-1.5 - 43 languages, 5x more efficient than Gemini Flash
- MAI-Thinking-1 - their first reasoning LLM
- MAI-Voice-2 - natural speech, 15+ new languages
- MAI-Voice-2-Flash - lightweight voice variant
- MAI-Code-1-Flash - tuned for GitHub Copilot

Microsoft is no longer just an OpenAI wrapper.
They're becoming a model lab.

[Watcher.Guru @WatcherGuru]

JUST IN: Claude AI developer Anthropic confidentially files for IPO.

[Abnormal AI @AbnormalAIX]

300+ AI tools in the game…

[Abnormal AI @AbnormalAIX]

Top AI Agents Events in June 2026 - Where AI builders & innovators meet!Mark your calendars:

• Microsoft Build 2026 → June 2-3, San Francisco
• Agentic AI Summit NY → June 4, New York
• The AI Summit London → June 10-11, London
• AI Engineer World’s Fair → June 29–July 2, San Francisco

Who’s attending which one?

[Abnormal AI @AbnormalAIX]

AI Attracts Nearly Two-Thirds of Global Investment Flows

Since the beginning of the year, AI companies have raised approximately $380 billion (accounting for 64% of total capital flows) through three channels:

- Investment bonds (IG bonds): approximately $140 billion (49% of total IG issuance)
- Venture capital (VC): approximately $220 billion (87% of total VC capital). This means that for every $10 invested in venture capital, nearly $9 goes to AI companies.
- High-yield bonds: approximately $21 billion (38% of total issuance).

Thus, not only crypto but also other sectors needing funding are facing fierce competition from AI in almost all fundraising channels.

[Abnormal AI @AbnormalAIX]

@ETH_L2_Insiders What are they planning to do with this stack?

[ClawUp @ClawUpAI]

150+ registrations.
100+ platform logins.
80 AI Agents created.

A quick look back at the ClawUp Workshop in Hanoi

Builders, developers, and AI enthusiasts came together to learn, experiment, and deploy their first AI Agents.

Thanks to everyone who joined us and made the event a success.

Ready to build your own AI Agent?
https://t.co/d1XFTffqLJ

[GOAT Network @GOATNetwork]

https://t.co/boWPT6Uico

[Abnormal AI @AbnormalAIX]

Codex CLI 0.135.0 is live ⚡
✦ Smarter codex doctor diagnostics
✦ Vim mode: text objects + better nav
✦ /status shows remote server info
✦ Python SDK Sandbox presets for thread/turn APIs
→ https://t.co/QWG9Z1IMMn

[Abnormal AI @AbnormalAIX]

Top AI Agent projects on X - May 2026 📊
🥇 @aixbt_agent - 1.05M followers (#1 AI on X)
🥈 @virtuals_io - 280K
🥉 @bittensor_ - 265K
4️⃣ @ai16zdao - 220K
5️⃣ @PAAL_AI - 110K
6️⃣ @griffaindotcom - 85K

$AIXBT is lapping the field. The gap between #1 and #2 is wider than the entire rest of the list combined.

[Abnormal AI @AbnormalAIX]

Anthropic released Claude Opus 4.8 - an incremental but meaningful upgrade over 4.7.

Key updates:
→ Effort control for users on Claude AI
→ Dynamic workflows in Claude Code for large-scale tasks
→ Fast mode now 3× cheaper, runs at 2.5× speed
Same price point. Worth watching.

[Abnormal AI @AbnormalAIX]

AI profitability doesn't come from having the most users.
It comes from embedding deepest into enterprise workflows.
Anthropic understood this early.
OpenAI is relearning it while burning $14B/year.
For builders working on AI stacks:
Enterprise use cases + workflow automation is still where value concentrates most over the next 2–3 years.
Build accordingly

[Abnormal AI @AbnormalAIX]

Why is Anthropic winning on unit economics while OpenAI is still burning cash?

3 things worth paying attention to:

1. Enterprise beats consumer on long-term profitability
OpenAI has 900M+ weekly users. Anthropic has 1,000+ enterprise customers spending $1M+/year — and that number doubled in just 2 months. Enterprise revenue is sticky, expands over time, renews consistently. Consumer subscriptions don't.
2. The efficiency gap is widening
Anthropic targets break-even by 2028. OpenAI pushed theirs to 2030 — and will burn 14x more cash before getting there. Same race. Completely different burn rates.
3. Distribution is the real moat
Claude runs on AWS, Google Cloud, AND Azure simultaneously. OpenAI is heavily dependent on Microsoft. When enterprise wants flexibility, Anthropic becomes the default.

[Abnormal AI @AbnormalAIX]

What if your AI could manage an entire team of agents on its own?

Claude Code's new dynamic workflows do exactly that - auto-generated orchestration + parallel subagents, all from one prompt.

Say the magic word: "workflow" 🪄

[Abnormal AI @AbnormalAIX]

@RWA_Terminal Exactly! Thank you for agreeing.
Infra is admitting they can't eat it all, the app layer is where the real value explodes

[Abnormal AI @AbnormalAIX]

The infra layer is literally telling us the app layer is a separate, massive opportunity.

When OpenAI & Anthropic pour billions into forward-deployed ventures, they're signaling: the next model won't solve it alone.

Like semis in the cloud supercycle - value migrates up the stack.

[a16z @a16z]

OpenAI and Anthropic are effectively telling the market they can't solve every problem with a generic AI coworker.

You don't pour billions into massive forward-deployed joint ventures if you think the next model release is going to take care of it.

In the cloud supercycle, semis led and software followed (and you didn't need Qualcomm or ARM to tell you the value was migrating up the stack).

In AI, the infra layer itself is telling us the application layer is a separate, massive opportunity they can't fully capture.

a16z's @joeschmidtiv on why the app layer isn't dead: https://t.co/84QN5Mj9T3

[ClawUp @ClawUpAI]

Have an idea for an AI Agent but not sure where to start?
Try https://t.co/45MyeMAnSB

Create your own agent in just a few clicks:
• Start with an idea
• Add tools and customize behavior
• Deploy instantly
• Let real users interact with it

No complicated setup.
No infrastructure headaches.
Just build, launch, and start experimenting.

[Abnormal AI @AbnormalAIX]

@ClawUpAI No-code/low-code agent builders are a strong trend.