Observed Signal · Jun 30, 2026 · Technical Release · Source: AI Secret · Impact: 4/5 · Sentiment: Neutral
In‑House Model Fever: Companies Build Their Own LLMs
The AI Secret newsletter reports a growing trend of companies building proprietary, narrowly specialized models instead of renting frontier models — exemplified by Base44's Base One and MyClaw.ai's upcoming MyClaw Pro. Meta released Brain2Qwerty v2, a non‑invasive brain‑activity decoder and open‑source research release that reached substantially higher word‑accuracy than prior non‑invasive systems. The newsletter also highlights AI-driven disruption in animation production reducing costs and jobs, and a revived Health and Location Data Protection Act introduced by Elizabeth Warren and Mary Gay Scanlon that would ban selling health and location data to brokers and explicitly cover user inputs to AI systems. A short TL;DR lists other industry moves (Anthropic, Google/Gemini, Adobe & Disney, OpenAI partnerships) underscoring rapid commercial, technical and regulatory shifts affecting AI adoption.
Major technical releases (Meta's Brain2Qwerty v2) and a revived federal privacy bill affecting AI data handling, combined with a broad industry trend toward proprietary in‑house models, have material implications for platform strategy, costs, data governance and regulation across advertising and marketing technology.
Track Wix Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Base44 — acquired by Wix last year for $80 million — put its first proprietary model, Base One, into production.
- MyClaw.ai said it will soon ship its own in‑house model, MyClaw Pro.
- Meta released Brain2Qwerty v2, a non‑invasive brain decoder trained on data from nine volunteers and open‑sourced the code.
- Senator Elizabeth Warren and Representative Mary Gay Scanlon revived the Health and Location Data Protection Act, banning sale of health and location data to brokers and covering user inputs to AI systems; the bill gives the FTC 180 days to write rules and $1 billion over ten years to enforce.
- Industry reporting (Bloomberg) says AI is materially cutting animation production costs, with some studios estimating 30–90% savings.
Connected Companies & Entities
10 Entities mapped“Base44, the app-building platform Wix bought for $80 million last year, just put its first proprietary model, Base One, into production....”
“Cursor moved first, Base44 and myclaw are following, and Shlomo expects every player with enough scale and data to do the same....”
“Cursor moved first, Base44 and myclaw are following, and Shlomo expects every player with enough scale and data to do the same....”
“The math also points straight at Lovable, the $500 million category leader, next....”
“Two Pixar veterans left for Google's DeepMind to make theirs....”
“OpenAI launched ChatGPT Health for medical records, and Anthropic shipped a HIPAA-ready Claude for Healthcare days later....”
“Adobe and Disney are using Foundry AI to design next-generation theme park rides....”
“Adobe and Disney are using Foundry AI to design next-generation theme park rides....”
“OpenAI launched ChatGPT Health for medical records, and Anthropic shipped a HIPAA-ready Claude for Healthcare days later....”
“According to Bloomberg, Animation is the first corner of Hollywood where AI is cutting real production, not just marketing....”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Frontier models and the case for owned, custom AI
A What’s Hot newsletter highlights a busy week of model releases from major labs (Meta, OpenAI, SpaceXAI) and spotlights Mira Murati’s Thinking Machines Lab and its mission to build multimodal, collaborative AI that organizations can own and customize. The author and their VC firm (boldstart) emphasize investing in teams that build proprietary models and data flywheels rather than only renting frontier models. The piece also references several related developments: Meta’s Muse Spark 1.1, OpenAI’s ChatGPT Work (powered by Codex and GPT-5.6), SpaceXAI’s Grok 4.5, Topos Bio’s Topos‑1, Netpreme’s X‑Mem MPU claims, Cloudflare’s Monetization Gateway waitlist (stablecoin settlement via x402), and the case for more U.S. open-weight models. Discussion topics include cost/performance tradeoffs, RL gains, local runnable frontier models, micropayments, and memory bandwidth bottlenecks in inference.
Enterprises Move to Own Their AI
The newsletter argues enterprises are shifting from renting frontier models to owning AI that encodes their data, workflows, and institutional knowledge. Two camps are emerging: firms that “own the context” (e.g., Palantir) and firms that “own the model” through open/post-trained models and routing. Recent platform moves underline the trend: Microsoft launched seven MAI models and promoted “Frontier Tuning,” NVIDIA shipped Nemotron 3 Ultra (and announced Cosmos 3), and startups like GeneralistAI raised a $400M round. The piece emphasizes that combining frontier capability with proprietary data lowers costs and increases control, and cites industry claims—such as Land O Lakes’ cost comparisons and Anthropic’s internal productivity metrics—as evidence the economics and technical paths for enterprise-controlled AI are maturing.
AI Flywheel: Cursor Sale, Open Models, Infra Booms
A Substack newsletter analyzes a surge of AI activity: Cursor’s reported sale to SpaceX for $60 billion highlights how application-layer companies that own workflows and data can outcompete model-makers. Open-source and Chinese models (e.g., Z.ai’s GLM-5.2) are closing performance gaps, prompting a multi-model enterprise world and increased post-training. AI infrastructure firms (Baseten) are raising large rounds, major M&A is active (Databricks buying Panther; Salesforce acquiring Fin/Intercom), and incumbents (Microsoft) are exploring lower-cost internal models like DeepSeek. The piece emphasizes the “data flywheel” — owning workflows to capture data, post-train models, and accelerate product-led growth — and calls out trends in edge GPUs, evals, agent-native startups, and AI security/infrastructure.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.