{{first_name | Leader}}, welcome back. Today’s updates:
See how to build secure MCP Auth for enterprise AI*
Google unveils faster, cheaper enterprise-grade AI compute chips
Google expands Gemini and launches global AI app-builder access
Tencent and Tsinghua introduce faster, token-free AI generation model
Tools, resources, and a prompt to quickly see which marketing or sales campaigns actually drove the pipeline. ⬇️
As AI agents connect to enterprise systems through MCP, secure authorization is essential. This guide explains how to implement OAuth 2.1 with PKCE, scopes, user consent, and token revocation to give agents scoped, auditable access without relying on API keys.
Learn how to design production-ready MCP auth for enterprise-grade AI systems.
Google’s new chips for faster, cheaper enterprise AI
Google rolled out its new Axion CPUs and Ironwood TPUs, built to cut training time and inference costs for large-scale enterprise workloads. Instead of the usual NVIDIA-led path, Google is pushing a “hypercomputer” setup that lets companies run bigger models with less wait time.
Google says these chips are optimized for multi-model workflows, meaning enterprises can run different AI systems in parallel without bottlenecks. That matters for teams juggling RAG, agents, and model orchestration. If Google delivers on cost and latency, this becomes one of the more practical infra upgrades of the year.
Google expands Gemini chatbot upgrades and Opal app builder
Google expanded Opal, its AI app-builder, to 160+ countries, meaning teams everywhere can now build internal AI tools without relying on engineering backlogs. Gemini also picked up new context-drawing and reasoning features, making it better at handling long documents and complex instructions.
The upgraded Opal stack means faster dashboards, lead triaging, and workflow tools without writing code. For enterprise sales and AI leaders, the new Gemini reasoning features offer better summaries, cleaner insights, and more reliable output.
Tencent and Tsinghua’s CALM model speeds up AI responses
Tencent and Tsinghua introduced CALM, a new modeling technique that skips token-by-token generation and instead predicts text in continuous chunks. This dramatically cuts response time and computing cost for chatbots, copilots, and assistants. If it scales, CALM could influence how enterprise tools deliver real-time AI interactions.
For enterprise product teams, faster inference means smoother customer support, quicker sales responses, and more reliable internal copilots. CALM also hints at where efficiency gains may come next.
Productivity Tools
Liminary - AI-superpowered second brain. Your insights, delivered when you need them. Never lose an idea. Get the Chrome Extension.
Firecrawl – Feed clean web data straight into your AI or analysis tools with minimal setup.
Wordtune - Helps you find plenty of wording alternatives to improve your text.
Twill – Offer employees digital health journeys that personalize care plans and track outcomes.
Get featured tomorrow: How do you use AI for business/personally? Interesting stories will be shared with 100K curious readers.
Useful Resources
A new Chinese AI model claims to outperform GPT-5 and Sonnet 4.5, sparking discussions on capability leadership.
Spatial supersensing is emerging as a core capability for multimodal AI, enabling richer environment understanding.
Top AI-agent sessions from SaaStr break down how to deploy agents that actually work in production.
AI-powered browsers are evolving into autonomous task assistants, not just search tools.
The best email-marketing automation tools for scaling communication workflows.
How long does it typically take your organization to move an AI initiative from concept to production
💰 Funding
FastBreak AI raised $40M to scale AI tools that improve decision-making and performance across professional and youth sports.
DealMaker raised $20M to expand its AI-driven capital-raising platform for modern fundraising teams.
💼 Roles in AI
🐦 Master Prompt Engineering
Campaign ROI Decoder
When to use this?
When you need to quickly see which marketing or sales campaigns actually drove the pipeline.
You are a performance strategist.
Review the campaign summary I’ll paste below and tell me:
Top 3 performing campaigns and what made them work.
Underperforming campaigns and why they missed target.
Patterns across audience, channel, or messaging.
3 concrete changes to increase ROI next quarter.
Output as:
| Campaign | Result | Reason | Recommended Change |Correct Input Style:
Here’s Q3 campaign data:
LinkedIn Ads – $80K spend, 420 leads, 22 opps, $1.2M pipeline.
Webinar Series – $30K spend, 260 leads, 6 opps, $300K pipeline.
Industry Event – $50K spend, 40 leads, 10 opps, $800K pipeline.”
P.S. Get more such prompts in the Prompting Playbook (free for you)
Q. Roughly what percentage of enterprise AI budgets in 2025 go toward data readiness (vs. model development)?
Answer: 97% of large enterprises have funding in place; only 18% are fully deployed as data quality and access top of the barrier list.
Stay curious, {{first_name | leaders}}
PS. If you missed yesterday’s issue, you can find it here.