INVA-AI-HERALD

AI NEWS

1. 📰 Zendesk Acquires Unleash to Add Enterprise Generative Search & Permissioned RAG

Zendesk acquired Unleash to embed permissioned RAG into employee workflows, establishing a new Israeli R&D hub. Integrating Unleash’s 70+ connectors and native Slack/Teams agents, the deal targets the Employee Experience (EX) market. Backed by Avigdor Willenz, the move accelerates secure, "in-the-flow" answers while strictly enforcing enterprise access controls. 

2. 🌍 Unframe Expands to Tel Aviv/Berlin with "Blueprint" Architecture

Unframe expanded to Tel Aviv and Berlin, backed by a $50M raise and led by founder Shay Levi. The "Blueprint" architecture bypasses requirements gathering for "Day 1" value in regulated EMEA markets. This outcome-based model counters the "Consulting Trap," delivering risk-free, compliant automation where enterprises pay only for verified results.

3. 🧪 Century Health & Tessel: AI for Muco-Obstructive COPD

Century Health partnered with Tessel Biosciences, backed by a $4.2M ARPA-H award, to target mucus-obstructive COPD. The collaboration combines deep phenotyping with organoid models to identify "mucus plug" patients missed by claims data. This accelerates trial design for non-responsive subpopulations, pinpointing specific "non-Type 2 inflammation" markers.

4. 🛡️ Commvault & Pinecone: Immutable Backups for Vector DBs

Commvault partnered with Pinecone to launch the industry's first immutable backup for vector databases, targeting 1H 2026 availability. The solution delivers Point-in-Time Recovery (PITR) and Cleanroom Recovery to protect RAG pipelines from "data poisoning." By deploying this for its internal "Arlie" assistant, Commvault ensures mission-critical inference workloads in banking meet strict data governance standards.

5. 🧰 United Rentals Scales "Manual Assist AI" to 4,000 Users

United Rentals scaled its "Manual Assist AI" to 4,000+ monthly users, utilizing Amazon Bedrock and OpenSearch to index 300,000+ pages of technical manuals. Championed by CIO Tony Leopold under the "Worksite Performance Solutions" strategy, the tool drastically cuts diagnostic times. This enterprise-wide deployment validates the operational ROI of GenAI, directly boosting utilization for its $22.8 billion equipment fleet.

6. ⚡ Tech Mahindra Deploys Gemini 2.5 via Orion Framework

Tech Mahindra integrated Google’s Gemini 2.5 multimodal models into its Orion Agentic AI framework, targeting 10% faster adoption for enterprises. The solution leverages Vertex AI and BigQuery connectors to deploy "governed agents" that reason across proprietary data. This architecture ensures strict air-gapped compliance while automating complex decision pipelines.

TRENDING AI NEWS

1. 💊 Takeda’s Zasocitinib Hits Phase 3 Goals

Takeda’s Phase 3 Latitude trials for zasocitinib achieved PASI 90 in >50% of patients and PASI 100 in ~30%, meeting all 44 secondary endpoints. This oral TYK2 inhibitor confirms a superior safety profile to older JAKs. With $6 billion peak sales potential, it challenges BMS’s Sotyktu, validating Takeda’s AI-driven pivot to precision dermatology.

2. 📈 Micron Surges 16% on AI Memory "Supercycle"

Micron shares surged 16% on an $18.7 billion forecast, driven by an "insatiable" HBM3E shortage. CEO Sanjay Mehrotra confirmed 2026 capacity is sold out, including initial HBM4 allocations. Raising FY2026 capex to $20 billion, the firm is aggressively ramping 1-gamma DRAM, signaling a multi-year "supercycle" that cements its dominance in AI data center infrastructure.

3. 🛑 UK Actors Vote 99% to Ban Digital Scans

In a landslide 99.6% indicative ballot, UK union Equity voted to ban unauthorized digital scans, rejecting Pact’s "perpetual rights" clauses. The vote demands strict consent frameworks for digital replicas, setting a global precedent. This industrial action threatens to disrupt UK film production unless studios guarantee residual payments and strictly limit AI-generated "performance cloning."

4. 🇨🇦 Canada Taps Coveo for "Sovereign AI" Search

 Canada signed an MOU with Coveo for "Sovereign AI," championed by Minister Joël Lightbound. The initiative distinguishes "Generative vs. Agentic," retrieving within "Protected B" perimeters. Using Model Context Protocol, it enforces the Directive on Automated Decision-Making, ensuring sensitive citizen data remains sovereign while automating public workflows.

5. 💻 Coder & GlobalLogic Partner for Agentic AI Dev

Coder and GlobalLogic formed a strategic alliance to deliver agentic AI environments to the Global 2000. Focusing on "Day 2" operations, the partnership combines Coder’s secure workspaces with GlobalLogic’s engineering scale. It provides a forensic audit layer for autonomous code, ensuring banking and healthcare clients meet strict Basel III and HIPAA compliance standards.

6. ⚡ Hut 8 Secures $7B Deal with Google Backstop

Hut 8 finalized a 15-year, $7 billion Triple-Net (NNN) lease for 245 MW of Anthropic infrastructure, operated by Fluidstack. Secured by a Google financial backstop, the deal generates $454M annual NOI and enables 85% loan-to-cost financing via J.P. Morgan and Goldman Sachs. Anchored by a 1,000 MW ROFO, this pivot to hyperscale compute ensures massive, de-risked revenue visibility.

📗 Book Summary: Seeing Like a State — How Certain Schemes to Improve the Human Condition Have Failed
Author: James C. Scott

🧠 Main Idea:
James C. Scott explains why many large, well-intentioned government and institutional plans fail. The core problem: states try to simplify complex human societies to make them “legible,” but this simplification destroys local knowledge and causes unintended harm.

🔍 Key Concepts:

• Legibility:
  Governments simplify society (maps, surnames, statistics, zoning) to manage it — but simplification removes vital nuance.

• High-Modernist Ideology:
  Blind faith in scientific planning and top-down control ignores human behavior and lived experience.

• Local Knowledge (Metis):
  Practical, ground-level knowledge developed by people over time is essential — and often ignored by planners.

• Why Big Plans Fail:
  When centralized power + rigid ideology + weak resistance combine, disasters follow (failed cities, farming schemes, social programs).

• Small-Scale Success:
  Systems work better when they evolve organically, adapt locally, and allow feedback and correction.

🌱 Core Lessons:

• Human societies are too complex for rigid top-down design.

• Local knowledge often outperforms expert planning.

• Good intentions are not enough — humility and adaptability matter.

• Systems should be flexible, participatory, and human-centered.                                                                                                                                                                                                                                                                                                                              AI TOOLS

• 🔮 CosmicGuide AI – AI astrologer that generates personalized readings and guidance instantly.

• 🎯 Oz – AI ad creator for image & video ads, including carousels, influencers, avatars, and presenters.

• 🪞 AI Face Shape Detector – Analyze a photo to identify your face shape with AI-powered precision.

• 🎨 ColoringStore – Transform ideas into printable, creative coloring pages using AI.

• 🤍 WhiteRank – Track and improve how AI search engines like ChatGPT, Gemini, and Perplexity mention and rank your brand.

• 📞 Voice Mate – AI-powered voicemail agent that answers missed calls, summarizes intent, schedules callbacks, and syncs with your calendar automatically.

• 🧹 Sliq – Automated AI data-cleaning tool that fixes formats, missing values, and schema issues to deliver analysis-ready data in minutes.

• 🛍️ Mujo AI – Generate complete ecommerce product visuals and conversion-focused copy from a single photo using AI design agents.

• 📖 Neural Novelist – An AI-powered novel writing assistant that helps fiction authors plan, write, and refine stories with creative guidance and structure support.

• 🔐 A2E Protocol – A declarative Agent-to-Execution protocol that enables secure AI agent automation by generating safe, controlled workflows without executing arbitrary code.