• 80/20 AI
  • Posts
  • Anthropic Introduces Artifacts in Claude Code

Anthropic Introduces Artifacts in Claude Code

Get your startup pitch torn apart — before an investor does it for you

Advertise here | 6-min Read

AI isn't just changing how work gets done. it's changing entire business models.

At Agency Advantage 2026, discover how leading agencies are leveraging AI agents, automation, and intelligent systems to increase output, improve margins, and stay competitive in a rapidly changing market. Join thousands of agency owners, marketers, and business leaders for two days of practical AI strategies and real-world insights.

June 30–July 1 • Free Virtual Event

Anthropic Introduces Artifacts in Claude Code

What’s Happening AI Today

1. Two of Google's top Gemini researchers are leaving for Anthropic — the talent drain is accelerating: Jonas Adler and Alexander Pritzel — both key contributors to Google's Gemini model — are moving to Anthropic, Bloomberg confirmed last night. Adler led Google's AI coding effort. Pritzel was central to how Gemini models are trained. This follows John Jumper, the Nobel Prize-winning AlphaFold architect, who joined Anthropic earlier this week. Google is losing the researchers who built its most competitive AI products — to the company it is trying to compete with. The exodus is now large enough that it constitutes a structural risk to Google's ability to close the gap with Claude.

2. OpenAI launched GPT-5.5-Cyber — its most capable cybersecurity model ever, 85.6% on CyberGym: OpenAI shipped the full version of GPT-5.5-Cyber as part of its Daybreak cybersecurity initiative. The model scored 85.6% on CyberGym — the highest score from any single model ever recorded on that benchmark — compared to 81.8% for standard GPT-5.5. It is not a public API. Access is gated through the Trusted Access for Cyber program to vetted organisations including Akamai, Cisco, Cloudflare, CrowdStrike, and Palo Alto Networks. The launch arrives the same week that the first confirmed autonomous LLM cyberattack was documented — timing that is either coincidental or very deliberately planned.

3. SpaceX's Colossus compute revenues now exceed $80 billion through 2029 — Anthropic is paying $45B of it: SpaceX signed a $6.3 billion compute lease with Reflection AI this week — the fourth major external tenant at its Colossus facilities. The full picture is now visible: Anthropic pays approximately $45 billion through mid-2029, Google pays approximately $30 billion, Cursor is included via acquisition, and Reflection AI adds $6.3 billion. SpaceX built Colossus to train Grok for xAI. It has become one of the largest commercial AI compute platforms in the world — and Elon Musk's competitors are paying the rent.

/

Get your startup pitch torn apart — before an investor does it for you

Prompt: You are a partner at a top-tier venture capital firm who has seen thousands of pitches and funded fewer than 1% of them. You do not give polite feedback. You give the feedback that saves founders from wasting two years on a story that does not hold up in a partner meeting.

Here is my pitch: [paste your deck, one-pager, or describe your startup — problem, solution, market, business model, traction, team, ask].

Tear it apart across five dimensions:

1. The problem — Is this a real, urgent, recurring problem that people are actively trying to solve right now? Or is it a vitamin disguised as a painkiller? What is the strongest argument that this problem does not need a new company to solve it?

2. The market — Is the addressable market large enough to build a venture-scale business? Walk through the bottom-up math. Where does the TAM claim fall apart under scrutiny? What would a skeptical analyst say about the size and timing?

3. The secret — What does this company believe that almost nobody else believes yet — and is there real evidence for it? Every fundable pitch has a non-obvious insight at its core. If I cannot name yours in one sentence, the pitch does not have one yet.

4. The competition — Who actually competes with this, including the incumbent solution of doing nothing or using a spreadsheet? What is the honest version of the competitive landscape — not the 2x2 matrix where you always win?

5. The kill shot — What is the single most likely reason this company fails? Not a list of risks. The one thing. Be specific enough that if I told it to the founder, they would either fix it immediately or realise the company should not exist.

End with a partner meeting verdict: pass, explore further, or invest — and the one sentence that explains why.

Do not soften the feedback to be kind. Soft feedback at this stage is the most expensive thing you can give a founder.

⚡QUICK HITS

Fable 5 jailbreak fully documented — the 120,000-character system prompt is now on GitHub. The technical details that triggered the government shutdown are publicly available.
OpenAI launches self-serve Ads Manager inside ChatGPT — advertisers can now buy placements directly inside ChatGPT conversations. The ad era of AI has officially begun.
OpenAI launches 3 real-time audio models — conversational agents, live translation, and transcription, all available via API today.
SanDisk stock up 5,000% in 2026 — trading at $1,980, with $42B in secured contracts. AI storage demand is reshaping the memory chip market faster than anyone predicted.
Meta buys $27B of Nebius GPU capacity — European-compliant compute for GDPR-sensitive AI training workloads, on NVIDIA's Vera Rubin platform.
Anthropic proposes a global AI pause — Daniela Amodei floated the idea publicly. Analysts are divided on whether it is genuine or strategic positioning ahead of the IPO.

AI news highlights

OpenAI and Broadcom unveiled an LLM-optimised inference chip — today — OpenAI's first custom silicon, built specifically for inference workloads. The compute stack just got a new layer.
Anthropic's updated privacy policy requires government-issued ID from July 8 — the most likely mechanism for restoring Fable 5 access to verified US citizens without fully lifting the export control directive
John Jumper — 2024 Nobel Prize in Chemistry, AlphaFold co-lead — joined Anthropic — the most credentialed scientific hire in AI history. Expected to lead Anthropic's AI for science initiatives.
xAI is raising $20B — $8B equity, $12B debt, NVIDIA at the core — Musk's AI lab is now capitalising at a scale that matches OpenAI and Anthropic's recent rounds.
Anthropic in early talks with Microsoft to run Claude on Maia 200 chips via Azure — CNBC confirmed. Claude inference on Microsoft's custom silicon would reduce Anthropic's SpaceX compute dependency.
Pentagon partnered with Shield AI to deploy autonomous drone swarms — Hivemind software on LUCAS loitering munitions. A single human operator manages the entire swarm. The autonomous weapons era is not hypothetical.
DuctGPT discovered rare-earth-free permanent magnets — using physics, not training data — Ames Laboratory's AI understands underlying physics and invents new materials. The model that does not exist in training data yet.
Gemini 3.5 Pro still in enterprise preview — 6 days left in June — Google's CEO committed to a June GA at I/O. Every day it does not ship, the credibility cost compounds.

Trending AI tools

Wispr Flow — Speak naturally in any app — it writes in your style, with auto-edits and command mode
Polygraph — AI agents with cross-repo visibility and persistent session memory — launched today
Framer Agents — Design, write, and organize your site with agents on the canvas — Claude Code and Codex compatible
Backgrind — Run AI agents over any app, sandboxed with worktrees and parallel sessions
Deep Work Plan — Repo-native specs and resumable tasks for longer agent refactors
Lyndi — AI nutrition coach that fits your lifestyle, not a generic diet plan
Zapier Agent — Automate multi-step workflows across 7,000+ apps with a single AI agent
Synthesia — Turn scripts into professional AI videos with realistic avatars — no camera needed

That’s a Wrap

SPONSOR US

Get your business in front of over 90k+ AI professionals

8020AI is the world’s #1 AI Newsletter, Read by 90k+ professionals from leading companies such as Google, OpenAI, Meta, and Microsoft.

We've assisted in promoting Over 500 AI-Related ProductsWill yours be the next?

What We Can Offer:

  • Launch an Advertising Campaign

  • Introduce New Product or Features

  • Other Business Cooperation

Or Email our founder Alamin at [email protected]

FEEDBACK

How was your experience with 8020AI today?

How was 8020AI today?

Login or Subscribe to participate in polls.

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.