Visual-based newsletter on business and tech

We explain the latest business, finance, and tech news with visuals and data.📊

All in one free newsletter that takes < 5 minutes to read. 🗞

Save time and become more informed today.👇

Subscribe with one click today

Today:

• Claude Investor: Future King?

• Stability AI CEO Exits

• iOS Teams with Anthropic

• FT Chatbot: Claude-Powered AI

• AI Investment Hits $13B Peak

Will "Claude Investor" DOMINATE the Future of Investment Research?" - AI Agent Proliferation Begins

Andrew Ng emphasizes the significance of AI agentic workflows, foreseeing their potential to drive significant AI advancements, even surpassing the impact of upcoming foundation model upgrades like GPT-5. He highlights the rapid progress in AI agents' capabilities, from basic concepts to complex tasks like investing. 

He also introduces a framework for categorizing AI design patterns, advocating for reflection, tool use, planning, and multi-agent collaboration. Open-source initiatives, like Claude Investor and Multimodal Gamer, further demonstrate AI's potential in finance and gaming. Despite advancements, challenges like latency persist, urging continuous refinement in AI development.

WATCH THE VIDEO ON YOUTUBE

Stability CEO resigns from generative AI company

Emad Mostaque, the CEO of Stability AI, has stepped down amidst controversies and internal issues, leaving the company in the hands of interim co-CEOs. Despite its past success and contributions to AI, Stability has faced challenges, including reports of corporate dysfunction. Mostaque emphasized his belief in decentralized AI as he departs. 

The company, known for its Stable Diffusion image generator, had raised significant funding previously. However, rumors of a sale and investor pressure led to Mostaque's resignation. This move comes amidst a broader trend of AI advancements, exemplified by the University of South Florida's plans to launch an AI-focused college.

AXIOS

Apple’s iOS future could also include Anthropic.

Apple's iOS future may involve integrating Anthropic's technology, alongside potential partnerships with OpenAI, Google, or Baidu for chatbot functionality. Mark Gurman's Power On for Bloomberg hints at Anthropic's involvement, proposing Apple may allow developers to integrate generative AI directly into iPhones. 

Additionally, iOS 18 may feature a revamped, customizable home screen, as detailed in the subscriber edition. This signals Apple's ongoing efforts to enhance user experience and leverage AI advancements within its ecosystem.

THE VERGE

Financial Times tests an AI chatbot trained on decades of its own articles

The Financial Times introduces Ask FT, an AI chatbot trained on its decades of articles, offering curated answers to subscribers' queries. Unlike generalized AI bots, Ask FT draws from the outlet's vast archive, providing well-sourced responses. Powered by Claude from Anthropic, the chatbot tackles current events and broader topics with citations for added credibility. 

While still in beta and limited to FT Professional subscribers, Ask FT demonstrates the publication's commitment to evolving its services and meeting user needs. With potential for deeper insights and refinement, it represents a strategic step forward for the 135-year-old institution.

THE VERGE

AI Spending to Surpass $13 Billion by 2028, Media Analysts Predict

Media analysts predict that AI spending will exceed $13 billion by 2028, with investments spread across analytics, development/delivery, and customer experiences. At a Series Mania presentation, research directors from Omdia and Plum Research emphasized AI as a tool rather than a threat, highlighting its potential to enhance efficiency and creativity. 

While acknowledging the limitations of certain AI applications, such as OpenAI's Sora, analysts foresee significant impacts in areas like storyboarding, previz, and postproduction. As AI becomes more integral to various industries, businesses face the challenge of integrating it effectively into existing workflows.

VARIETY

🧠RESEARCH

ReNoise: Real Image Inversion Through Iterative Noising

"ReNoise: Real Image Inversion Through Iterative Noising" tackles the challenge of faithfully reversing real images into the domain of text-guided diffusion models. The method introduces iterative renoising, refining predictions along the diffusion trajectory without increasing operations. It enhances reconstruction accuracy and preserves editability, proving effective in both speed and precision.

Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition

"Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition" presents a novel approach, CMD, to enhance video diffusion models' efficiency without compromising quality. CMD decomposes videos into content frames and low-dimensional motion latents, leveraging pretrained image diffusion models. This innovative design significantly improves generation quality and reduces computational costs, outperforming prior methods in both speed and quality.

MyVLM: Personalizing VLMs for User-Specific Queries

"MyVLM: Personalizing VLMs for User-Specific Queries" addresses the limitation of large-scale vision-language models (VLMs) in understanding user-specific concepts. The method introduces external concept heads to enable VLMs to recognize user-provided concepts, facilitating personalized image captioning and visual question-answering. Experimental results demonstrate generalization to unseen concepts while maintaining model performance on unrelated inputs.

IDAdapter: Learning Mixed Features for Tuning-Free Personalization of Text-to-Image Models

"IDAdapter: Learning Mixed Features for Tuning-Free Personalization of Text-to-Image Models" introduces a novel approach, IDAdapter, for personalized portrait generation without test-time fine-tuning or multiple input images. By integrating personalized concepts through textual and visual injections and a face identity loss, IDAdapter enriches identity-related content details from multiple reference images, resulting in diverse styles, expressions, and angles in generated images while preserving identity fidelity.

HyperLLaVA: Dynamic Visual and Language Expert Tuning for Multimodal Large Language Models

"HyperLLaVA: Dynamic Visual and Language Expert Tuning for Multimodal Large Language Models" introduces a novel approach for enhancing Multimodal Large Language Models (MLLMs) performance. By incorporating adaptive tuning of the projector and language model parameters, alongside dynamic visual and language experts derived from HyperNetworks, HyperLLaVA achieves significant improvements over existing benchmarks, addressing static tuning constraints across diverse multimodal tasks. The method demonstrates superior performance on various MLLM benchmarks, outperforming previous approaches like LLaVA.

🛠️TOP TOOLS

ChatGPT at Work - Ultimate guide to revolutionizing your workflow with 100 prompt ideas to help you unleash the power of generative AI.

Just words - Maximize learnings with real time copy optimization, versioning and no-code experimentation

Simply News -Run entirely by a team of AIs. Updated daily.

Super Rag - Super performant RAG pipelines for AI apps. 

Tattoos AI - AI-powered tattoo generator

📲SOCIAL MEDIA

🗞️MORE NEWS

Nvidia could be primed to be the next AWS

Nvidia and Amazon Web Services (AWS) share similar growth stories, both benefiting from unexpected opportunities in cloud and AI markets. Nvidia's revenue surged to $22.1 billion in Q4 2024, primarily from data center business. While AWS faces slower growth, Nvidia anticipates sustained expansion, propelled by AI demand. Yet, competition looms. TECHCRUNCH

Accenture’s $1 billion LearnVantage platform tackles the growing AI skills gap

Accenture launched LearnVantage, a $1 billion AI-powered learning platform to address the skills gap in emerging technologies. It offers personalized learning paths, leveraging content from Accenture, Udacity, and tech giants like AWS and Google Cloud. With industry-specific training, it aims to help businesses stay competitive amid technological advancements. VENTUREBEAT

Apple researchers explore dropping “Siri” phrase & listening with AI instead

Apple researchers explore using AI to detect when users talk to devices like iPhones without needing a trigger phrase like "Hey Siri." They trained a model using speech and background noise data, aiming to enhance user experience. However, concerns about privacy and transparency arise if the trigger phrase is removed. Apple's AI advancements signal competition with tech giants. MIT

NHS AI test spots tiny cancers missed by doctors

NHS trials AI tool "Mia" alongside doctors, detecting tiny breast cancers missed by humans. Mia flagged 11 cases out of 10,000 mammograms, potentially aiding early detection. Patients like Barbara benefited, receiving less invasive treatment. While promising, AI isn't perfect; it may overlook existing cysts and lacks access to patient history. Further research and validation are needed to maximize its potential in cancer diagnostics. BBC

A Booming Nvidia Supplier Says AI Costs Need to Drop

Micron, a key supplier for Nvidia's AI chips, faces uncertainty over sustained demand due to high AI costs. Senior VP Raj Narasimhan admits challenges in predicting demand shifts. Micron, sold out through 2025, reflects booming AI chip demand, but questions linger about long-term sustainability amid cost concerns. THE INFORMATION

Ready or not, AI chatbots are here to help with Gen Z’s mental health struggles

The emerging digital health industry is offering hundreds of free apps, including Earkick and Woebot, to address the mental health crisis among teens and young adults. These AI chatbots are designed to provide comforting and sympathetic statements, breathing exercises, and stress-management tips, but they do not claim to diagnose or treat medical conditions. Some experts argue that chatbots could help with less severe mental and emotional problems, but others express concerns about their limitations, including the ability to recognize emergency situations. AP NEWS

Nvidia’s AI ambitions in medicine and health care are becoming clear

Nvidia's push into healthcare is paying off, with new AI tools showcased at its 2024 conference. Partnerships with Johnson & Johnson and GE Healthcare underscore its commitment. AI accelerates drug discovery, cutting costs and time. Despite concerns, AI's potential in healthcare is undeniable, marking a lucrative venture for Nvidia's future growth. CNBC