- TheNeural
- Posts
- This Week in AI with The Neural
This Week in AI with The Neural
Claude Powers Finance, Google Calls, Robots Cut, Grok Goes Gov
TECHIES, ASSEMBLE!
AI Highlights
Claude goes Wall Street with a new financial services suite
Robots now cut gallbladders—autonomously
ChatGPT Agent launches, doing multi-step tasks solo
Google Search can now call local businesses for you
Meta's open-source party may be coming to an end
Anthropic and DoD partner for responsible AI
ElevenLabs clones DJ Sam Feldt’s voice, globally
NVIDIA shows how to train an LLM in one weekend
AI INDUSTRY
OpenAI and UK Government Announce Strategic Partnership
The UK signs an MoU with OpenAI to boost national productivity with AI. The focus: safer AI integration in public services and infrastructure. It’s a formal step toward weaving foundational models into government workflows.
Grok for Government Announced
xAI launches Grok for Government, bringing AI tools to federal and state agencies. With a $200M DoD contract in hand, the suite offers secure access to Grok 4, Deep Search, and Tool Use to accelerate public sector transformation.
Anthropic and US DoD Partner on Responsible AI
Anthropic has partnered with the Department of Defense in a $200M deal to develop secure and steerable AI. It’s focused on critical infrastructure and prototyping frontier applications while upholding democratic values in defense tech.
CDAO Collaborates with Frontier AI Companies
The DoD’s CDAO has signed on OpenAI, Anthropic, Cohere, and Meta to help test and deploy frontier AI responsibly. The goal: mission-ready models that align with U.S. values and can scale in national security settings.
Zoho's Zia LLM Stack and Agent Studio Launch
Zoho debuts its own LLM stack and Agent Studio, taking control over its AI destiny. It’s moving away from third-party providers to ensure sovereignty, cost-efficiency, and domain-specific optimization for enterprise clients.
China’s EngineAI Raises $139M for Humanoid Robots
EngineAI raised $139M to scale its humanoid robots—built for logistics, manual labor, and elder care. China is placing big bets on robotic automation, vying with the West in the race to humanoid deployment.
Meta’s Free AI Model Days Are Numbered
Meta’s open-source generosity may soon fade. With defense contracts and infra costs mounting, Bloomberg warns Meta could shift toward closed models—threatening the accessibility of frontier AI for startups and researchers.
Diligent Robotics Hits 300K Pharmacy Deliveries with Moxi
Moxi, the hospital robot, has completed over 300,000 pharmacy deliveries—quietly becoming a backbone of medical logistics. With secure med handling and low infrastructure needs, it’s scaling into broader hospital roles.
Elix + LINC Launch Commercial AI Drug Discovery Platform
Elix and LINC launched a federated AI platform that allows 16 pharma companies to jointly train models—without sharing sensitive data. It improves discovery accuracy and aims to become Japan’s standard for AI-driven drug R&D.
Mistral Launches Open-Source Audio Model
Image Source: Mistral.AI
Mistral released Voxtral, an open-source audio model suite with top-tier accuracy and multilingual Q&A. It's designed for companies that want affordable, open, voice intelligence that rivals the best proprietary offerings.
GitLab Duo AI Agent Enters Public Beta
GitLab Duo AI enters beta with agents that handle dev tasks across code, tests, and security scans. It also supports “Flows”—multi-agent sequences that tackle large tasks end-to-end. It’s a step toward agentic DevOps.
‘HOW TO’ WITH AI
Train a Reasoning-Capable LLM in One Weekend (NeMo)
NVIDIA shows how to train a reasoning LLM on a single GPU in 48 hours. Using open data and LoRA adapters, it toggles between chat and advanced logic modes. A compact guide proves that powerful AI isn’t just for giants.
AI Around Us
Claude for Financial Services
Claude is entering Wall Street. Anthropic's new suite unifies market feeds, Excel data, and internal reports—all under Claude's guidance. Its capacity is built for finance-specific workloads and speeds up processes like modeling and due diligence, while preserving privacy by not training on client data.
AI-Enabled Robot Performs Gallbladder Surgery
A surgical robot guided by an AI model similar to ChatGPT has successfully performed gallbladder removal in lab settings. Developed by Johns Hopkins, this robot self-corrects and reached 100% success rate—signaling a near-future where autonomous surgery could become reality.
Firefly Adds Video & Sound Effects AI Tools
Adobe Firefly just got a creative upgrade. It now generates sound effects and animates avatars from text. With tools like Style Presets and Keyframe Cropping, creators can produce polished, dynamic video content at speed—while keeping commercial licensing watertight.
Google Search Gets Agentic Upgrades
Google is making Search far more than a search engine. Now, it can make phone calls to check business hours, run deep research queries, and return cited reports—all powered by Gemini 2.5. Think of it as your personal research and scheduling assistant.
ChatGPT Agent Launch
OpenAI’s ChatGPT Agents are here: mini AI-workers that handle multi-step tasks across tools like browsers and code interpreters. They can plan, reason, and act on your behalf—no code needed. It’s a shift from chat-based Q&A to action-based delegation.
Voice AI Reaches Fans Anywhere: Sam Feldt's Use Case
DJ Sam Feldt uses ElevenLabs to stay connected with fans—without needing a studio. The AI clones his voice, translates it into multiple languages, and delivers studio-quality content anywhere. It’s efficient, global, and supports his carbon-neutral tour ethos.
Gemini 2.5: Conversational Image Segmentation
Gemini 2.5 takes image understanding to a new level. It doesn’t just detect objects—it understands abstract prompts like "the person holding an umbrella.” This allows developers to ask deeper visual questions for editing, safety checks, or scene analysis.
Le Chat Gets Deep Research Mode & Voice Recognition
Le Chat, Mistral’s assistant, now offers voice interactions and a deep research mode that generates structured reports. Its multilingual reasoning and project folders make it a powerful tool for organizing ideas, conducting research, and editing images—all in-chat.
New AI Image Editor by MIT
Image Source: MIT
MIT has redefined image generation—without needing traditional image generators. Their method manipulates image tokens directly, allowing edits and creations with less compute. It’s fast, lean, and could set new standards for AI image creation workflows.
Claude Helps Validate SaaS Ideas in Minutes
A Reddit user validated a SaaS idea in 10 minutes using Claude. By scraping real user complaints from Reddit and G2, the AI pinpointed a market gap—leading to a cold outreach tool now making $2.3k/month. It’s startup validation, AI-style.
AI POV
OpenAI: AI as a Source of Empowerment
OpenAI doubles down on its vision: AI as a tool for everyone. From agents to personalization, the company argues that AI will amplify individual productivity—not replace it—and can democratize creative and economic opportunity.
Google’s AI Strategy Starves Publishers
Futurism warns that Google’s AI-powered summaries are siphoning traffic from news outlets. With AI Overviews delivering answers directly, users skip clicks—and publishers lose ad revenue. Accuracy and media survival both hang in the balance.
You're Invited: ACTIVATE 2025
India’s AI future takes the mic.
Join thinkers, builders, and decision-makers at ACTIVATE 2025 — a one-day summit by The Neural on converting AI ambition into real-world action.
📍 Bangalore | 📅 August 8, 2025
From banking to biotech, discover how AI is reshaping industries from the inside out.