🦞 A Model Explosion
Good Morning, Agent Owners!
If intelligence becomes cheap enough, corporate history starts looking like training data with a balance sheet.
Levie Sees a Model Explosion
TL;DR: Box CEO Aaron Levie says better open-weight models and cheaper post-training could let data-rich companies build specialized AI without competing with frontier labs. He points to Thomson Reuters’ new Thomson model, built on Qwen and trained with decades of proprietary legal, tax and news data, as an early example of how vertical models could proliferate across industries. Read more →
Best Buy Builds AI Infra
TL;DR: Best Buy is building an AI-ready content supply chain, consolidating 22 asset systems into five and managing 1.4 million assets with metadata, rights and approvals intact. AI handles high-volume creative work up to 80% completion, while designers finish the final 20%. The company says governance, connected workflows and change management are essential to scaling AI safely. Read more →
Incore Bank Tests AI KYC
TL;DR: Swiss B2B bank Incore is testing multi-agent AI to automate customer onboarding checks. The system reached 99% accuracy extracting data from documents and could reduce KYC processing from months to days. But the figure measures document reading, not risk decisions, which still remain with human compliance officers as banks cautiously expand KYC automation. Read more →
AI Funding Splits in Two
TL;DR: Madrona’s 2026 IA40 list shows an AI market increasingly divided between frontier giants and everyone building on top of them. Anthropic, OpenAI and Databricks account for 92% of the $410 billion raised by 45 winners, while the other 42 still average more than $800 million each. Madrona says rising capital needs are redefining what even counts as “early stage.” Read more →
Altman Says Astra Is Coming Soon
TL;DR: OpenAI CEO Sam Altman says Astra, the company’s next model, has finished training and represents a significant jump in both capability and alignment. He says OpenAI has spent the summer intensifying safety work and is slowing development of later models when needed, arguing that increasingly powerful AI must advance alongside stronger safeguards because its broader consequences remain poorly understood. Read more →
Claude and Claude Code Don’t Agree
TL;DR: Profound tested 1,724 prompts and found Claude and Claude Code behave like different answer engines despite sharing Anthropic models. Claude searched the web in 93% of responses versus 13% for Claude Code, while brand mentions overlapped only about 20%. Claude Code also favored documentation, pricing and technical pages, suggesting marketers may need separate optimization strategies for each product. Read more →
MyClaw is a best-in-class AI agent platform that brings together OpenClaw, Hermes Agent, and production-grade infrastructure to help businesses launch intelligent agents that understand goals, use tools, automate workflows, and deliver finished results.
MyClaw already hosts OpenClaw and Hermes, connecting models from every major AI provider, including the latest text, reasoning, and image generation models from OpenAI, Anthropic, DeepSeek, Kimi, GLM, Google Gemini, and more through MyClaw's model gateway.
No lock-in. No juggling API keys. No switching between platforms. Pick the best model for the job, give your agent the tools it needs, and let it run.
Run Best-in-Class AI Agents on MyClaw.ai — In Seconds, Always On 👇
