This week on No-BS AI Briefing:

- Google pushes back Gemini 3.5 Pro launch, even scrapping an original base model after it missed internal performance targets, especially in coding and tool-use. This highlights significant vendor roadmap risk and strengthens the near-term position of competitors like OpenAI and Anthropic.

- Moonshot AI unveils Kimi K3, a massive 2.8 trillion-parameter open-weight model. Reports suggest it's competitive with Claude Fable 5 on front-end coding and GPT-5.6 Sol on the Artificial Analysis Intelligence Index, offering builders new options for cost control and flexibility.

Deep Dive: We go deep on Google’s Gemini delay, discussing why even top-tier vendors can miss targets, the strategic implications for startups and enterprises, and the importance of maintaining model fallbacks and diversified dependencies.

Practical Takeaway: Audit your vendor roadmap dependencies. Learn how to map your critical features to AI models and vendors, and brainstorm Plan B alternatives in under 30 minutes to build more resilient AI products.

Connect with Vikash Sharma on LinkedIn for questions or topics you want covered!

Send us Fan Mail

Support the show

This week on No-BS AI Briefing, we’re talking about a landmark EU decision that could reshape mobile AI, new tools for reliable enterprise agents, and the strategic importance of power infrastructure for AI.

(1) **Google Gemini Enterprise adds real-time web grounding with citations:** Google launched Grounding with Parallel Web Search for Gemini Enterprise, integrating real-time search with verifiable, cited web sources to combat hallucinations. This creates a "grounding-as-a-service" pattern, simplifying reliability for builders. [Source: Google Developers Blog]

(2) **Ledger unveils hardware-backed safety stack for AI agents:** Ledger's open-source toolkit blocks autonomous blockchain transactions without physical device confirmation and secures AI credentials with OpenPGP encryption. This offers a concrete safety layer for finance, DeFi, and regulated workflows. [Source: Crypto Economy]

(3) **EU orders Google to open Android APIs and share anonymized search data:** Binding rules mandate access for rival AI agents to Android voice-activation and background task APIs, plus anonymized search data sharing by January 2027. This is a landmark move opening a dominant platform to competitive AI. [Source: US News & World Report]

(4) **Bunkerhill Health raises $25M Series B for hospital AI agents:** Led by Khosla Ventures, this funding validates enterprise agent deployments in healthcare operations and clinical algorithms, offering a template for regulated verticals. [Source: Fortune]

(5) **Nebius secures $1.7B for AI cloud power infrastructure:** Industrial Development Funding and Oaktree are deploying Bloom Energy fuel cells for Nebius’s AI cloud, highlighting energy as a core scaling bottleneck for AI and making power infrastructure a strategic moat. [Source: TradingView / PR Newswire]

**Deep Dive:** The EU's mandate to open Android APIs and share search data is a game-changer, creating new distribution channels and data access for rival AI agents. We discuss the strategic implications for startups, big tech, and application builders, alongside potential risks.

**Practical Takeaway:** Ask your team: "How would our product change if Android voice and background APIs opened to us, *and* we had access to anonymized search data, free and reliably?” Use a quick brainstorming session to identify 2-3 radical new use cases and assess their effort vs. impact.

---

Send us Fan Mail

Support the show

This week on No-BS AI Briefing, we unpack Google Cloud's significant move to integrate Anthropic's Claude models directly into its Agent Platform, offering enterprise controls and model choice. We also look at xAI's new Zero Data Retention feature for Grok Build CLI, a game-changer for IP and compliance. Plus, learn why private equity firms like Clearlake are betting big on AI, signaling deep demand for practical AI solutions. For our deep dive, we break down why Google Cloud's multi-model strategy is crucial for builders. And for your practical takeaway, we'll walk you through a quick team exercise to brainstorm new product ideas unlocked by these changes.

Send us Fan Mail

Support the show

This week, Meta's infrastructure VP, Barak Yagour, issued a stark warning: enterprise systems have "maybe 20 months" to adapt to the demands of AI agents, signaling a fundamental architectural shift. We delve into why agent workloads break traditional infrastructure and what this means for builders.

We also cover Microsoft's impressive feat of patching 570 vulnerabilities in a single day using AI, Arist's launch of voice-based AI diagnostic agents for enterprises, Intrinsic Power's funding to optimize AI data center electricity, and a new Google.org RFI for AI in K-12 assessment.

Join Vikash Sharma as he unpacks these high-signal stories, offering practical insights and a concrete takeaway to help you audit your infrastructure for the coming age of AI agents.

Send us Fan Mail

Support the show

Welcome to the No-BS AI Briefing, where we cut through the hype to bring you critical AI news for founders, builders, and engineers.

In this episode:

* **Anthropic expands Claude Cowork to iOS, Android, and web:** Discover how cross-device, human-in-the-loop agents are setting new UX and compliance standards for enterprise AI workflows.

* **Senator Warner floats the first federal bill on agentic AI:** We deep-dive into the "AI AGENT Act" discussion draft, exploring its implications for fiduciary duties, data privacy, transparency, kill-switches, and interoperability—and why you need to start designing for auditability now.

* **Anthropic extends free Fable 5 access:** Learn why this time-sensitive opportunity is crucial for benchmarking Fable 5 against OpenAI's GPT-5.6 Sol to inform your production model choices.

* **UK unveils £45M music industry plan, flags AI as a challenge:** Understand the emerging landscape of sector-specific AI policy, rights, and licensing for creative AI tools.

**Deep Dive:** Senator Warner’s AI AGENT Act – This is a game-changer for the U.S. regulatory landscape, shifting the focus to how autonomous agents are governed. We break down what it means for product design, engineering, and strategic planning, and offer a no-BS take on what's real versus speculative.

**Practical Takeaway:** Don't miss the chance to benchmark Anthropic's Fable 5 against OpenAI's Sol this week. We give you a quick, actionable plan to evaluate these frontier models for your specific use cases before the free access window closes.

Stay ahead without the noise. Follow for more concise, opinionated briefings!

Send us Fan Mail

Support the show

In this No-BS AI Briefing, Vikash breaks down OpenAI's safety restructuring and its implications for builders. Get the lowdown on Grok 4.5's new public API pricing and coding focus, Google's mandatory AI ad labels for compliance, and Saudi Arabia's agentic AI, INSAIGHTS.

Send us Fan Mail

Support the show

In this episode of No-BS AI Briefing, Vikash unpacks key AI developments from July 11, 2026.

* **Ethereum Foundation uses AI agents to find critical node bug (CoinDesk):** Learn how AI is being deployed for high-stakes security, identifying a remote crash flaw in Ethereum's node software, and why human validation remains crucial.

* **AI scribes cut physician burnout by 74% in study (KevinMD):** Discover the real-world impact of ambient AI on talent retention and operational efficiency in high-compliance sectors.

* **Phia: $185M AI shopping assistant launch (Fortune):** A look at a major consumer AI launch and the immediate challenges around ethical data practices and transparent consent.

* **Mistral unveils Robostral Navigate (single-camera robot nav) (Mistral AI):** Explore how new 8B-parameter models are reducing hardware costs and accelerating deployment for logistics and service robotics.

* **Meta pauses Instagram AI image feature amid consent backlash (The New York Times):** A critical lesson for builders on the risks of opt-out consent models when dealing with user data and likeness.

**Deep Dive:** We go deeper into the Ethereum AI security story, discussing why AI-assisted auditing is a game-changer for distributed systems, who should care, and how builders can approach this new capability without falling for the hype.

**Practical Takeaway:** Vikash suggests a hands-on experiment: run AI-assisted code review on a non-critical repository to understand its practical signal-to-noise ratio in your own workflow.

Don't miss out on concise, actionable AI insights for founders, product leaders, and engineers. Follow "No-BS AI Briefing" and connect with Vikash Sharma on LinkedIn for more.

Send us Fan Mail

Support the show

This week, we explore the tangible shift of agentic AI from experimental pilots to core operational infrastructure. Qualcomm's acquisition of Modular and partnership with Hugging Face are set to redefine how AI workloads are dynamically routed between devices and the cloud, emphasizing autonomous task placement based on latency and cost. IBM, in collaboration with Riyadh Air, is pioneering an "AI-native airline" built on modular watsonx agent skills, offering a blueprint for regulated industries. We also highlight Hidden Brains' success with voice agents, demonstrating significant ROI in real-world production environments. Finally, we discuss new US legislative proposals targeting harmful chatbot design for minors, signaling increased scrutiny on AI product UX and the need for emotional dependence risk assessments. Our deep dive focuses on the strategic implications of the AI-native enterprise model for builders, and we offer a practical exercise: modeling your own workflows as modular "skills" to prepare for the agentic future.

Send us Fan Mail

Support the show

Foreman: Self-hosted LLM gateway with cost-aware routing (GitHub, July 8, 2026)

Makoto: Integrity hook for Claude Code (GitHub, July 8, 2026)

Send us Fan Mail

Support the show

In this episode of No-BS AI Briefing, we cover Microsoft's significant move to deprecate the Azure OpenAI Assistants API, giving builders a hard deadline to migrate their agent workflows. We'll unpack what this means for your agent strategy and the broader implications of platform lock-in.

We also dive into:

- **AMD's new Ryzen AI Halo Developer Kit**: Making on-device AI inference more accessible for developers.

- **Swarm**: A native Swift framework for building powerful, stateful multi-agent AI workflows.

- **Illinois AI Safety Measures Act**: A new state law imposing compliance requirements and hefty fines on AI developers.

- **ShareClean**: A local-first CLI tool for automating secret redaction in logs and configs.

Stay ahead of the curve with practical insights and actionable takeaways.

Connect with Vikash Sharma on LinkedIn for questions and topics.

Send us Fan Mail

Support the show