SunBrief#75: $1.8B Startup With Almost No Team

Welcome to the SunBrief

Today in SunBrief 🌞

• 1,000+ ChatGPT Prompts from Superhuman AI

• AI Helps Two Brothers Build a $1.8B Startup With Almost No Team

• OpenAI Raises $122B at $852B Valuation to Scale AI Infrastructure

• Stock Updates

• Anthropic leaks Claude Code by accident

• AI Highlights of the Week

• Too Important to Miss

Supercharge Your Productivity With 1,000+ ChatGPT Prompts from Superhuman AI

ChatGPT is insanely powerful.
But most people waste 90% of its potential by using it like Google.
Master AI and work 10X faster with 1,000+ proven ChatGPT prompts.
Sign up for Superhuman AI to get:

• 1,000+ ready-to-use prompts to solve problems in minutes instead of hours—tested & used by 1M+ professionals

• Superhuman AI newsletter (3 min daily) so you keep learning new AI tools & tutorials to stay ahead in your career—the prompts are just the beginning

AI helped two brothers build a $1.8B company

One entrepreneur used a stack of AI tools to run coding, marketing, and support with almost no staff

Matthew Gallagher built Medvi using AI to handle coding, ads, content, and support. With only his brother as a full-time hire, the startup quickly scaled to hundreds of millions in revenue and is now on track for $1.8B in annual sales.

Key Points:

• Tiny team, huge scale: Gallagher launched Medvi with AI tools and minimal human help, later hiring only his brother as the company’s main employee.

• Fast growth curve: Medvi reportedly hit 300 customers in month one, added 1,000+ in month two, then reached $401M sales in 2025 and is now on pace for $1.8B.

• AI ran core operations: He used AI to build the site, write code, create ads, handle customer service, and analyse business performance—then outsourced doctors, pharmacy, shipping, and compliance to partners.

• Real risks showed up: The AI support system sometimes hallucinated prices and products, and a small website mistake once caused major downtime because there was no team to catch it.

• Profitability + expansion: Medvi claimed strong net margins and is expanding into men’s health, meal plans, and more categories—still keeping headcount extremely low.

Why It Matters:
This shows what AI native companies can look like: fewer employees, faster execution, and massive scale without large teams. But it also comes with tradeoffs, less redundancy, higher operational risk, and systems that can be efficient yet surprisingly fragile.

OpenAI Raises $122B at $852B Valuation to Scale AI Infrastructure

Funding backs compute expansion, enterprise growth, and a unified “AI superapp” strategy

OpenAI has raised $122B at an $852B valuation, positioning itself as core AI infrastructure and highlighting compute as the key advantage driving model quality, adoption, and cost efficiency at scale.

Key Points:

• Massive funding round: OpenAI says it secured $122B in committed capital and expanded its revolving credit facility to ~$4.7B (undrawn).

• Big-name backers: The round was anchored by Amazon, NVIDIA, and SoftBank, with continued participation from Microsoft and many major institutions.

• Consumer scale claims: OpenAI says ChatGPT has 900M+ weekly active users and 50M+ subscribers, with usage and revenue accelerating rapidly.

• Compute-first strategy: OpenAI highlights a multi-provider, multi-chip approach across Microsoft, Oracle, AWS, CoreWeave, Google Cloud, and silicon across NVIDIA, AMD, Trainium, Cerebras, plus a chip effort with Broadcom.

• “AI superapp” plan: OpenAI says it’s unifying ChatGPT, Codex, browsing, and agent features into a single agent-first product to turn consumer distribution into workplace adoption.

Why It Matters:
If OpenAI’s numbers hold, this round signals a new phase where AI leaders raise infrastructure-scale capital to lock in compute, push agentic products into everyday workflows, and compete on distribution, not just raw model quality.

Stock Updates

Anthropic leaks Claude Code by accident

An npm packaging mistake published internal Claude Code code paths, raising IP and security concerns

Anthropic accidentally shipped a large JavaScript sourcemap file in the public npm package for @anthropic-ai/claude-code (v2.1.88), which developers say revealed a massive chunk of Claude Code’s internal TypeScript code and design patterns. Anthropic said it was a release packaging issue caused by human error, not a breach, and that no customer data or credentials were exposed.

Key Points:

• What leaked: A ~59.8MB .map file included in the npm release exposed internal implementation details and was quickly mirrored and analyzed by the community.

• Why it matters technically: The leak reportedly reveals how Claude Code manages long-running agent sessions (memory/indexing strategies and orchestration logic), giving competitors a clearer path to replicate “agent harness” behavior.

• Anthropic’s response: The company confirmed the incident, framed it as a packaging mistake, and said it’s rolling out prevention measures.

• Security concern around the ecosystem: Separate from the sourcemap issue, multiple outlets flagged a broader npm supply-chain risk environment at the same time, meaning users should be extra cautious with dependencies and updates.

• Axios compromise advisory: Microsoft published guidance on an Axios npm supply chain compromise, including recommended mitigation steps for affected environments.

Why It Matters:
Even without data leaks, exposed tooling can aid competitors and attackers. Teams should review dependencies, check recent updates, and follow security guidance amid npm supply chain risks.

AI Highlights of the Week

• OpenAI Acquires Tech Talk Show TBPN

  OpenAI has acquired TBPN, the popular founder-led business and tech talk show, making it the company’s first media acquisition.
  The show will keep its own brand and editorial independence, while OpenAI uses its strong comms and marketing reach to help explain AI’s growing impact on daily life.

• Claude just shut the door on OpenClaw

  Anthropic is ending the ability to use Claude Pro and Max subscriptions with third-party agent tools like OpenClaw, pushing those users toward API or extra usage billing instead.

  The move reflects growing compute pressure and tighter capacity control, but it also risks frustrating power users who relied on cheaper subscription access for heavy agent workflows.

• Google Launches Gemma 4 Open Models

  Google has launched Gemma 4, a new family of open models built from Gemini 3 research and technology.
  The lineup ranges from compact models for mobile and edge devices to larger versions designed for powerful local AI on PCs, with support for reasoning, multimodal tasks, and 140 languages.

• Microsoft Unveils 3 New MAI Models

  Microsoft has introduced MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2, now available in Foundry and MAI Playground.
  The new models promise better quality, faster speeds, and lower costs across speech, voice, and image generation.

Too Important to Miss

Perplexity’s Computer Now Handles U.S. Tax Work

SpaceX and Amazon Clash Over Satellite Collision Risks

Google Finally Lets U.S Users Change Their Gmail Address

Smartphone Price Hikes Look Inevitable in 2026

Last Week’s Poll Result

• Should AI be allowed to model and simulate human brain activity at scale?

  Yes, for science and medicine → 38.46%
  Maybe, with strict regulation → 30.77%
  No, too risky ethically → 30.77%

• Is Apple making a smart move by turning Siri into a platform for multiple AI models?

  Yes, best of all worlds → 44.44%
  Maybe, but Apple is losing control → 27.78%
  No, they should build their own AI → 27.78%

  

• What’s the biggest risk from leaking “personal” emails of high-profile officials?

  Indirect national security risks → 44.83%
  Future targeting and blackmail → 37.93%
  Exposure of contacts and networks → 10.34%
  Reputational damage → 6.90%

Feedback

Know someone who may be interested?