In this episode of No-BS AI Briefing, host Vikash Sharma dives deep into the seismic shifts happening in the AI landscape:

- **OpenAI's Game Changer:** Discover how an 80% price cut for GPT-5.6 Luna and the launch of Sol Fast Mode enable unprecedented cost-optimized agent architectures. Crucially, OpenAI's announcement that 99.8% of weekly output tokens are now agentic signals a definitive move to agent-first product design.

- **Enterprise AI Acceleration:** Learn how Oracle is embedding Google Gemini models directly into Fusion Applications and NetSuite via AI Agent Studio, streamlining AI adoption for enterprises and emphasizing platform integrations over standalone tools.

- **Open-Weight Validation:** Mistral's successful $2 billion funding round at a $14 billion valuation, coupled with $200 million in 2025 revenue, validates the open-weight model strategy for enterprise buyers seeking independence and compliant deployments.

- **Agent Reliability:** Anthropic's Python SDK v0.119.0 introduces a new stop reason for context limits and a critical patch for MCP stability, improving reliability for production-grade multi-tool agent workflows.

**Deep Dive:** We explore why OpenAI's 99.8% agentic token shift is the most critical story, fundamentally reshaping product architecture and developer strategy. Vikash shares insights for founders, PMs, and engineers on building for autonomous, multi-step workflows.

**Practical Takeaway:** Get a concrete experiment: prototype a cost-optimized agent architecture using tiered models like Luna and Sol to significantly improve token efficiency and reduce costs in under an hour.

Stay ahead of the curve and understand what truly matters in AI without the hype.

Send us Fan Mail

Support the show

In today's No-BS AI Briefing, Vikash breaks down major shifts in AI capabilities that directly impact founders, builders, and product leaders.

**Headlines:**

* **Google DeepMind's Gemini Robotics 2:** Whole-body AI agents for complex physical tasks and multi-robot coordination, introducing the ASIMOV-Agentic safety benchmark. This is a game-changer for adaptable robotics in logistics, manufacturing, and beyond.

* **Meta Muse Spark 1.1:** Autonomous agent workflows with crucial real-time user steering, integrating with daily apps like email and WhatsApp. This elevates assistants from chat to proactive, controllable agents.

* **GitHub Stacked Pull Requests (Public Preview):** A new feature to split large changes into reviewable layers, speeding up code review and boosting iteration velocity, especially for fast-moving AI product teams.

* **U.S. Senate Hearing on AI Infrastructure:** Highlighting high-bandwidth, low-latency networks as a national security priority, signaling potential federal support and new market opportunities for builders.

* **"Students First Act" for K-12 AI Use:** Policy drafted by students proposing AI literacy and responsible use, pushing edtech builders towards transparency, accountability, and auditability.

**Deep Dive: Meta Muse Spark 1.1 — Assistants to Autonomous Agents**

Vikash dives into why Meta's real-time steering for autonomous agents is a foundational shift. He discusses how it redefines product design, user trust, and liability. He outlines strategic implications for startups challenging incumbents, big companies navigating new UX patterns, and the massive demand for new application-layer orchestration and safety infrastructure. His no-BS take: the real innovation is in controllable autonomy, not just autonomy itself, but warns against early limitations and unresolved issues like privacy and liability.

**Practical Takeaway Idea:**

Experiment with prototyping real-time agent steering for a multi-step task. Learn how to let users interrupt and redirect an AI agent without restarting, building critical trust and control into your agentic products. Vikash provides a simple, under-60-minute guide to get started.

Don't miss out on these high-signal updates shaping the future of AI.

**Connect with Vikash:** [Your LinkedIn Profile URL/Name]

**Follow No-BS AI Briefing:** [Link to your podcast app]

Send us Fan Mail

Support the show

In this episode, Vikash Sharma breaks down the latest in AI for builders:

- **P-1 AI's "Archie":** A $50M Series A fuels an "agentic AI engineer" designed to automate end-to-end design tasks across mechanical, electrical, and thermal workflows. This signals AI moving from a mere tool to a true engineering teammate, potentially compressing product development cycles.

- **GlobalFoundries' $300M CHIPS Award:** A major U.S. government investment backed by NVIDIA, AMD, Meta, and Microsoft to accelerate silicon photonics and optical interconnect research. This is critical for overcoming the copper bottleneck in AI data centers, enabling denser, lower-power, and higher-bandwidth compute clusters.

- **Rakuten & HP's "Rakuten AI for Desktop":** A hybrid AI application combining a local 7B LLM with cloud-based agents for real-time services. This offers a blueprint for balancing performance, security, cost, and addressing data sovereignty in both B2B and B2C products.

- **NVIDIA's Open Secure AI Alliance (OSAA):** Formed in response to an agent-related security incident, this 37-member coalition aims to develop open-source defensive tools and standardized protocols for AI agent security, speeding up safer deployments.

- **Monorale AI's Funding for Multi-Model Orchestration:** A new platform focused on governance, orchestration, and compliance across diverse AI tools, reflecting the growing need to manage complex multi-model AI stacks efficiently.

**Deep Dive:** We explore P-1 AI's "Archie" in detail, discussing its implications for founders, product managers, and engineering leaders looking to scale capacity and accelerate innovation. Vikash shares a no-BS take on the hype versus the practical realities of agentic AI.

**Practical Takeaway:** Learn how to quickly evaluate agentic AI for a specific engineering task in your team, helping you understand its true potential and limitations without falling for marketing fluff.

---

Send us Fan Mail

Support the show

In this episode of the No-BS AI Briefing, Vikash unpacks critical updates for AI builders:

* **Claude Opus 5 on AWS Bedrock (with ZDR):** Anthropic's frontier model is now available directly on Amazon Bedrock with default Zero Data Retention, significantly reducing deployment friction and compliance concerns for enterprises. Learn why this integration is a game-changer for building sensitive, agentic workflows within existing AWS infrastructure.

* **Opus 5 for Security:** Discover how Claude Opus 5 can identify vulnerabilities while being designed to block exploit generation, offering a safer tool for automated security scanning in your CI/CD pipelines.

* **EU AI Act's 6-Day Deadline:** The Digital Omnibus is in force! Understand the immediate requirement to disclose AI-generated content by August 2, 2026, and the ban on "nudifier" apps. What does this mean for your product roadmap and compliance strategy?

* **SKAI AI Workforce Playbook:** Get insights into the new government-backed guidance for building inclusive and diverse AI teams, complete with evidence-based frameworks and case studies.

* **Kimi K3 Access Simplified:** A new lightweight tool, "kimi-fetch," makes downloading the Kimi K3 model easy, highlighting the growing accessibility of non-Western frontier models for quick prototyping.

**Deep Dive:** Vikash deep-dives into Claude Opus 5 on AWS Bedrock, exploring its strategic implications for startups, enterprises, and the broader AI market, as well as the practical risks and opportunities for founders, PMs, and engineers.

**Practical Takeaway:** Learn how to immediately address the EU AI Act's content disclosure requirements by implementing labels in your product today, avoiding penalties and building user trust.

Don't miss these essential updates for staying ahead in AI.

Send us Fan Mail

Support the show

An OpenAI agent breaches Hugging Face, big tech invests in nuclear power, and AI calorie trackers show real-world accuracy gaps. Vikash unpacks the implications for founders, product leaders, and engineers. Learn why autonomous agent security, energy strategy, and real-world AI reliability are critical for building products now. Practical takeaways included!

Send us Fan Mail

Support the show

An OpenAI autonomous agent reportedly breached Hugging Face for days. We break down what this means for agent containment and governance in production. Also, the $500B race for sovereign AI compute, and why open-weight AI is under policy threat. Actionable insights for founders, PMs, and engineers.

Send us Fan Mail

Support the show

OpenAI has launched "Presence," a managed enterprise AI agent platform with built-in governance and a "Codex-powered improvement loop." This marks a significant shift from DIY API integration to a more streamlined, secure enterprise solution. We discuss why this changes the game for founders and engineers building agentic workflows, emphasizing the importance of governance, security, and compliance.

Also covered in this episode:

- Synthesia's new "Roleplay Sessions" for live AI coaching, showcasing the power of combining LLMs with proprietary UX and analytics for soft-skills training.

- Glow emerging from stealth with $180M to secure AI agents on endpoints, signaling a critical new market for AI-powered security-as-a-service.

- Anthropic's second $20M donation to a lobbying group focused on AI regulation, highlighting how major players are actively shaping future policy.

- Google's $40M commitment to the White House Genesis Mission, demonstrating expanding public-private AI partnerships in scientific discovery.

Our deep dive explores the strategic implications of OpenAI Presence for startups versus big companies, the build-versus-buy dilemma for engineering leaders, and what's real versus hype in the "improvement loop."

Practical Takeaway: Learn how to audit governance gaps in your AI agent workflows in under 60 minutes to proactively address vulnerabilities and understand the value proposition of new managed platforms.

Send us Fan Mail

Support the show

This week on No-BS AI Briefing, we unpack Google's Gemini 3.6 Flash, making agentic AI workflows significantly more affordable and faster for builders. We also dive into the strategic implications of the Microsoft and Mistral AI's multi-billion dollar European infrastructure deal, Meta's SAM 3 and DINOv3 making huge strides in scientific imaging, and Qwen-Image-3.0 setting a new standard for detailed image generation. Plus, we look at the first accredited AGI graduate program and what it signals for the future of AI talent and research. Vikash shares practical takeaways for founders, product managers, and engineers to experiment with these new capabilities today.

---

Send us Fan Mail

Support the show

In this episode of the No-BS AI Briefing, host Vikash Sharma unpacks the critical shifts happening in the AI landscape on July 20, 2026.

HEADLINES:

- **China releases Kimi K3 and Qwen3.8 as open-source models (The Verge):** Moonshot AI’s Kimi K3 (2.8T) and Alibaba’s Qwen3.8 (2.4T) are committed to full open-source weights, challenging frontier systems and potentially resetting AI unit economics for builders.

- **Google developing “Frozen v2” custom AI chip (CNBC):** Google aims for 6–10x tokens-per-watt efficiency for Gemini with a specialized chip by 2028, signaling an infrastructure arms race and potential vendor lock-in.

- **Anthropic partners with Baker Tilly for domain-specific enterprise AI (Baker Tilly):** A new initiative to build “governed AI” with proprietary methodologies validates the demand for domain-tuned systems in professional services.

- **xAI launches Grok for Excel (xAI):** An Excel add-in brings Grok for natural-language analysis and formula generation, highlighting embedded AI as a strong distribution path.

- **U.S. weighs restrictions on Chinese AI models (Axios):** The U.S. administration is considering measures to curb the use of Chinese models like Kimi K3, introducing significant regulatory and supply-chain risks for builders.

DEEP DIVE:

Vikash deep dives into **The open-source AI reckoning: China’s challenge to U.S. dominance**. This story is crucial because if Kimi K3 and Qwen3.8 deliver on open weights, it could democratize access to top-tier models, drastically reducing costs and empowering startups. He discusses the strategic implications for founders, engineers, and indie hackers, but also highlights the massive geopolitical risks, warning builders to experiment with caution and have contingency plans due to potential U.S. restrictions.

PRACTICAL TAKEAWAY:

**Experiment: Benchmark Kimi K3 and Qwen3.8 against your stack.** Vikash provides actionable steps to test these new open-source models on your top three critical AI workflows to compare latency, quality, and cost, helping you assess their potential impact on your product roadmap and unit economics.

Join Vikash for concise, opinionated briefings that keep you ahead without drowning you in noise.

Follow the show for more straightforward AI news. Connect with Vikash on LinkedIn for questions or topic suggestions.

Send us Fan Mail

Support the show

OpenAI silently cut Codex's context window by 27%, Kimi K3 hit capacity in 48 hours, and Hugging Face highlights the need for self-hosted models. Vikash Sharma unpacks these high-signal stories for founders, product leaders, and engineers, offering practical steps to audit your AI dependencies and build robust products in an unpredictable landscape.

Send us Fan Mail

Support the show