- The Rundown AI
- Posts
- Vibe coding startup's $80M speedrun
Vibe coding startup's $80M speedrun
PLUS: OpenAI prepares for bioweapon risks with new model safeguards
Zach Mink & Rowan Cheung
June 20, 2025
Good morning, AI enthusiasts. Top AI minds have predicted AI will soon enable solo unicorn businesses — and a solo-bootstrapped vibe coding platform just provided some early evidence.
With Maor Shlomo’s Base44 going from zero to 250k users and an $80M acquisition in just six months, the AI-fueled entrepreneurship movement is just getting started.
P.S. — Our next live workshop is today at 4 PM EST. After the session, you’ll be able to confidently build and run your own automation agents using n8n! RSVP here.
In today’s AI rundown:
Solo-owned vibe coding startup sells for $80M
OpenAI prepares for bioweapon risks
How to build a smart AI orchestrator for projects
Stanford study: What workers want from AI
4 new AI tools & 4 job opportunities
LATEST DEVELOPMENTS
BASE44
Image source: CTech
The Rundown: Developer Maor Shlomo just sold his vibe coding startup, Base44, to Wix for $80M in cash, after bootstrapping it to 250k users and making monthly profits of $189,000 in just six months.
The details:
Shlomo said Base44 grew to 10k users within three weeks via word-of-mouth, enabling non-programmers to build apps with natural language prompts.
The Israeli developer bootstrapped the company and is the only shareholder, with his eight employees receiving $25M in bonuses as part of the acquisition.
Wix plans to integrate Base44 into its tools to help users build apps, with Shlomo calling the platform the “best possible partner” to continue scaling.
Shlomo initially started Base44 as a side project and launched in January, quickly landing partnerships with major companies like eToro and Similarweb.
Why it matters: AI leaders like Sam Altman and Dario Amodei have predicted AI advances could soon lead to solo unicorn startups — and Base44’s sale is strong proof of both the changing economics of software creation and the massive demand for the “vibe coding” movement that has been one of the biggest headlines of the AI wave.
TOGETHER WITH ARTISAN
The Rundown: Artisan is a fully automated outbound platform powered by Ava, your AI BDR. She handles prospecting, enrichment, outreach, and booking — delivering a constant pipeline of qualified leads straight to your inbox.
With Ava, you get:
300M+ B2B leads, including specialized sources like E-Commerce & Local Business
Deep enrichment across 10+ premium data providers
End-to-end deliverability and inbox management
Automated multi-channel outreach and personalized messaging that actually converts
Scale outbound without scaling headcount — book a demo to see what Ava can unlock for your team.
OPENAI
Image source: Reve / The Rundown
The Rundown: OpenAI just published a new blog detailing the safety measures being taken in preparation for the dangerous thresholds for biological weapon creation that the company expects its next generation of models could reach.
The details:
OpenAI anticipates successors to its o3 reasoning model will trigger the “high risk” status under its preparedness framework for biological threats.
Mitigations include training models to refuse harmful requests, deploying always-on systems to detect suspicious activity, and advanced red-teaming.
The company is also planning a July biodefense summit with government researchers and NGOs to discuss risks, countermeasures, and research.
The move follows similar safety measures from Anthropic, which recently activated stricter protocols for its Claude 4 family release.
Why it matters: The same capabilities that could unlock scientific breakthroughs will also enable some extremely dangerous stuff in the wrong hands — and this next step up in models will take the stakes to a serious new level. While the additional safeguards and proactive moves are positive, we’re quickly about to enter very unknown territory.
AI TRAINING
The Rundown: In this tutorial, you will learn how to create an intelligent system that automatically chooses between fast and powerful AI models based on task complexity, optimizing speed for your personal projects.
Step-by-step:
Get your OpenAI API key from the Platform dashboard
Install the OpenAI library: “!pip install openai” and set up your client connection.
Test a basic API call to understand the structure: “client.chat.completions.create(model="o3", messages=[...])”
Build your orchestrator function that rates task complexity (0.0-1.0) and auto selects between GPT-4o (fast) and o3 (powerful) based on your threshold.
Pro tip: Adjust your threshold value, e.g., lower (0.3) for quality priority and higher (0.7) for faster responses. You can access the complete notebook here!
PRESENTED BY INNOVATING WITH AI
The Rundown: 900+ founders have already joined Innovating with AI’s flagship program, and the next wave kicks off this week. With market demand set to 8x, now’s the moment to level up your consulting game.
In this cohort, you’ll learn:
The tools and frameworks to find clients and deliver top-notch services
A six-month roadmap for growing a six-figure consultancy
How alumni booked their first paying client in as little as three days
Enrollment ends June 27 — Click here to request access to The AI Consultancy Project.
AI RESEARCH
Image source: University of Stanford
The Rundown: Stanford surveyed 1,500 workers to map their AI automation desires, revealing critical mismatches between what employees want and what the tech industry is building, and finding workers prefer partnership over replacement for tasks.
The details:
The study revealed disconnects between desires and current AI development, with 41% of YC startups focused on areas workers considered low priority.
The results showed workers primarily want to automate low-value, repetitive jobs like scheduling and data entry to free up time for more important work.
The researchers also created a “Human Agency Scale,” finding nearly half of occupations preferred equal human-AI partnership over full automation.
Arts/media professionals show the strongest resistance to automation, with only 17% of creative tasks receiving positive ratings from workers.
Why it matters: Agents are set to play a massive role in the future of work, and insights into how workers portray AI collaboration and automation can go a long way toward finding areas to free up productivity. But as the tech surpasses humans in nearly every area, full-scale levels of automation may come whether it’s wanted or not.
QUICK HITS
🎥 V1 - Midjourney’s new image-to-video model
🎨 Higgsfield Canvas - Image editing model with advanced inpainting features
🗣️ Google Search Live - Talk, listen, and explore in real time with AI Mode
⚙️ VibeCode - A mobile app that builds mobile apps
Meta is in negotiations to hire AI investors Nat Friedman and Daniel Gross (also a co-founder of Ilya Sustkever’s SSI) to join Alexandr Wang’s superintelligence division.
OpenAI is reportedly planning to “scale back” its work with data startup Scale AI following its deal with Meta, joining Google, xAI, and Microsoft.
Perplexity launched new video generation capabilities, enabling users to generate Veo 3 videos with audio on social media by tagging the @AskPerplexity account.
OpenAI rolled out ChatGPT Record, a new feature allowing the assistant to capture, summarize, and transcribe audio from meetings and brainstorms.
Nvidia-backed SandboxAQ released SAIR, a dataset of 5.2M synthetic protein-drug molecules to train AI models for drug discovery.
Mass General Brigham researchers developed AI-CAC, a tool that reads chest CT scans to quickly spot calcium deposits that indicate potential heart disease.
COMMUNITY
Join our next workshop today at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. By the end of the workshop, you’ll be able to confidently build and run your own automation agents using n8n.
RSVP here. Not a member? Join The Rundown University on a 14-day free trial.
That's it for today!Before you go we’d love to know what you thought of today's newsletter to help us improve The Rundown experience for you. |
See you soon,
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
Reply