Your weekly briefing on all things AI

AI adoption faces one major hurdle—how to get users comfortable with new tools. But what if the key to breaking through isn't building standalone apps, but embedding AI into platforms people are already using every day? Enter Slack’s latest update, which allows users to integrate AI agents directly into their workflows. This isn’t just a product update; it’s a blueprint for how companies like Asana, Adobe, and others can win the AI adoption game.

Let’s dive in:

> Slack Gets Smarter: Introducing Game-Changing AI Agents

> Province-Wide AI-Powered Health Assistant Deployed in China

> Black Forest Labs Image Generator Raises at a $1B Valuation

> Spinach AI: The Virtual Teammate You’ve Been Waiting For

> Prompt Tip: How to Master ChatGPT’s Latest Model

From Chat to Action: Slack’s New AI Agents Revolutionizing Productivity

Slack’s latest update lets users integrate AI agents directly into their workflows, giving companies like Asana, Adobe, and others a prime opportunity to bring their AI tools into the hands of everyday users. Instead of relying on separate apps or systems, these AI agents can now work within Slack’s familiar interface, streamlining tasks like project management, content creation, and customer support without disrupting existing workflows. By embedding AI into a platform millions already use daily, this move promises to make advanced technology more accessible and less intimidating for businesses.

From a strategic standpoint, this is a distribution goldmine. Slack becomes the key entry point for cutting-edge AI adoption, while its partners—like Asana—get immediate access to a vast user base that’s already comfortable within the ecosystem. For tech companies looking to break through the AI adoption barrier, seamless integration into platforms like Slack is a smarter play than trying to push standalone products. The takeaway? In the AI arms race, success won’t just come from building innovative tools—it’ll come from embedding those tools where users already are. (Link)

OpenAI Shifts to For-Profit Model as CTO Steps Down

OpenAI is making major changes, moving from non-profit control to become a for-profit benefit corporation aimed at attracting more investors. The non-profit will still hold a minority stake, and CEO Sam Altman is set to receive equity as the company’s valuation is projected to reach $150 billion. Alongside this shift, OpenAI’s longtime CTO, Mira Murati, has unexpectedly left the company, signaling further leadership changes as OpenAI evolves to meet both investor expectations and its ambitious AI goals. (Link)

More Stories

• Websites can charge AI bots for scraping with Cloudflare’s new marketplace. (Link)

• Salesforce's Marc Benioff shares Steve Jobs’ impact on Agentforce AI. (Link)

• Ex-Apple Designer, Jonny Ive, Joins Forces With OpenAI. (Link)

• Drip Capital Leverages AI to Revolutionize Cross-Border Trade Operations. (Link)

• Jensen Huang and Marc Benioff highlight massive potential for agentic AI. (Link)

• Amazon introduces Project Amelia, a seller-focused AI assistant. (Link)

• Sam Altman steps down from OpenAI’s safety committee. (Link)

• Oracle’s Larry Ellison predicts AI will track every move in the future. (Link)

• 1X releases generative models to train robots. (Link)

• EA explores AI to enhance user-generated content. (Link)

• Neuralink secures FDA breakthrough for "Blindsight" implant. (Link)

• AI models are advancing to PhD-level reasoning, tools usage, and agent capabilities. (Link)

• UAE’s G42 and Nvidia collaborate on AI-driven climate technology. (Link)

• LinkedIn trains its AI tool with user-generated content unless they opt-out. (Link)

|| Healthcare

Zhejiang Launches AI-Powered Health Assistant for Improved Patient Care

Zhejiang has become the first province in China to introduce a province-wide AI-powered digital health assistant, Anzhener (Angel), in collaboration with Alipay. The AI system, now integrated into the Zhejiang Provincial People's Hospital, offers cloud-based medical escort services, health consultations, and real-time navigation via an Alipay mini-app. Serving 90,000 patients last month, Anzhener has helped streamline hospital visits, reduced workload at information desks by 50%, and achieved a 98% satisfaction rate. With over 1,000 medical institutions adopting the technology, the system is expected to expand to cover inpatient and emergency services soon. (Link)

|| Real Estate

AI Set to Revolutionize Australian Real Estate Industry

Barry Plant, an Australian real estate agency, is preparing to launch an AI-powered assistant named Grace, which will handle tasks such as communicating with potential buyers, responding to inquiries, and creating property listings. Set to debut in October, Grace will analyze property data and offer 24/7 assistance online. While Barry Plant is embracing AI to enhance customer experience, the agency remains cautious in its adoption. The Real Estate Buyers Agent Association of Australia (REBAA), however, warns that despite AI advancements, physical property inspections remain essential for comprehensive due diligence. (Link)

|| Finance

Hong Kong to Issue AI Policy for Financial Sector Amid Tech Tensions

The Hong Kong government is preparing its first policy statement on artificial intelligence in finance, aiming to guide ethical AI use and provide principles for its application across areas like trading, investment banking, and cryptocurrencies. The Financial Services and Treasury Bureau is currently drafting the guidelines, with industry feedback still being considered. Set for release around late October during Fintech Week, the statement is expected to clarify the city's stance on AI while addressing regulatory uncertainties. This move comes as the city faces challenges accessing major AI services due to the US-China tech conflict, positioning Hong Kong to strengthen its role as a financial hub amidst growing competition from Singapore. (Link)

• Black Forest Labs raises $100M at a $1B valuation for image generation. (Link)

• Datamaran secures $33M for ESG governance and risk management. (Link)

• PictorLabs raises $30M in Series B to enhance AI-powered virtual staining. (Link)

• 11x.ai gathers $24M, led by Benchmark, for AI digital employees. (Link)

• Fal.ai lands $23M from a16z and others for media-generating AI models. (Link)

• AI notetaker Fathom raises $17M. (Link)

• Ferrum Health secures $16M in Series A to drive AI adoption in healthcare. (Link)

• Inbolt raises €15M to boost AI for industrial robots. (Link)

• AI coding assistant Supermaven attracts $60M from OpenAI and Perplexity co-founders. (Link)

• Rep.ai secures $7.5M to launch digital twin sales representatives. (Link)

• Cudis raises $5M for blockchain-based AI wearable health ring. (Link)

Spinach AI Can Now Participate in Meetings

Backed by Y Combinator, Zoom, and Atlassian, Spinach AI is an AI note-taker for meetings, that has evolved to now participate in the meetings it takes notes for. Laying out the agenda and answering questions about previous meetings it attended are just a couple of the impressive new features. The tool could probably benefit from a voice upgrade, but for now, the robotic sounding agent is a welcomed addition, effortlessly joining meetings, capturing key insights, and preparing follow-up emails—all while seamlessly integrating with your existing tools. With its ability to support 100 languages, Spinach AI breaks down barriers, empowering teams worldwide to work smarter together.

As Gartner predicts that 80% of project management tasks will be automated by AI by 2030, Spinach AI is already paving the way for this future, allowing you to focus on what truly matters.

More tools to try:

• Epsilla: an all-in-one platform for building customizable AI agents using private data, designed for professionals and enterprises with ease, security, and scalability.

• Invoice Data Extraction: OpenAI-powered invoice extraction transforms mixed formats into custom spreadsheets with speed and precision.

• Dynamiq: streamline AI development with rapid prototyping, testing, and deployment, all within your infrastructure.

• TailorTask: transform tedious tasks in just 10 minutes—simply describe the job, and an AI agent will take care of it.

• Flarecut: converts written content into short videos featuring AI-generated voices and automatic subtitles.

• Fal.ai: a generative media platform empowering developers to create the next generation of creative content with lightning-fast inference.

• Zivy: transforms chaos into clarity with AI that prioritizes essential messages for engineering and product leads, keeping your focus razor-sharp.

• Langflow: redefines AI development with a focus on simplicity and flexibility for both experimentation and real-world applications.

• Replit: accelerates software development, enabling quick building, sharing, and deployment.

• Polymet AI: accelerate design and coding with AI for rapid prototyping and faster shipping of production-ready solutions.

• Meco: transforms your newsletters into a dedicated reading space and declutters your inbox in moments.

• Writecream: optimizes your writing process with AI, generating compelling blog articles and product descriptions in seconds.

• Plotline: revolutionizes your app's flexibility with AI, allowing rapid onboarding and adoption shifts without the usual engineering constraints.

• Superflex: accelerates front-end development, enabling you to build features ten times faster.

• Eagle Hunt: scans your code for vulnerabilities before committing, acting as your AI Code Security Assistant.

• AI Phone: phone app that transforms calls with live translation, breaking down language barriers for seamless communication anywhere in the world.

Mastering the New ChatGPT-o1 with OpenAI’s Prompting Guide

OpenAI released a new model, available to paid ChatGPT users and through their API, and it works differently than previous LLMs. Since this model performs chain-of-thought prompting internally, the optimal prompts for the new ChatGPT differ significantly. Here are some useful tips and examples to help you get the most out of it:

1. Keep Prompts Simple and Direct

• other models: "Can you provide a detailed explanation of why exercise is important for health, including physical and mental benefits?" (Overly complex and confusing.)

• o1: "List three benefits of exercise." (Clear and concise for easy understanding.)

2. Avoid Chain-of-Thought Prompts

• other models: "Can you think step by step about the causes of climate change?" (Unnecessary steps complicate the request.)

• o1: "What are the main causes of climate change?" (Directly asks for information without extra reasoning.)

3. Use Delimiters for Clarity

• other models:
  "What are the benefits of exercise, like improving heart health and reducing stress?" (This prompt is too wordy and lacks structure, making it harder for the model to identify specific benefits.)

• o1:

  "What are the benefits of exercise regarding:

  • Heart health

  • Stress reduction

  • Better sleep"

  (The list format clearly separates each benefit, making it easy for the model to identify and respond to them.)

4. Limit Additional Context to What’s Necessary

• Other models: "According to this report, solar energy and several other renewable sources reduce carbon emissions significantly, along with various case studies and statistics." (Excessive context can overwhelm the model.)

• o1: "Solar energy reduces carbon emissions significantly." (Focuses on essential information needed for the response.)