Published Date:

Report: 53% of Work Can Be Done by AI

That's how much the average employee spends on busywork!
By

​Microsoft introduces new multi-agent system Magnetic-One​

The future of having your truly “AI team” is here!

Microsoft researchers just announced Magnetic-One, a generalist multi-agent system that can carry out complex, multi-step tasks by itself across the web and local files.

Magnetic-One uses a lead agent called Orchestrator agent, which is similar to Salesforces’ Atlas reasoning engine in Agentforce, plans how to conduct a task, directs the other four specialized agents, monitors progress, and adjusts plans as needed to make sure “the AI team” is completing a given task.

  • WebSurfer: Manages tasks on Chromium-based web browsers.
  • FileSurfer: Reads local files and navigates directories.
  • Coder: Writes code, analyzes information, and creates new artifacts.
  • Computer Terminal: Provides a console shell for executing code and installing programming libraries.

The system is available for testing within Microsoft’s AutoGen framework and can be used with different reasoning LLMs.

Besides that huge release, Microsoft also:

​Claude 3.5 Haiku Sparks Debate​

Anthropic released its new lightweight LLM Claude 3.5 Haiku, which is:

  • Surpassed its predecessor Claude 3 Haiku and even the large Claude 3 Opus on intelligence and safety benchmarks.
  • Performing faster and better on coding suggestions, data extraction and labeling, and content moderation.
  • Having longer text outputs and a more recent knowledge cutoff (July 2024).

But the upgraded version also raised some eyebrows as:

  • Despite being positioned as a "budget" model, the cost has increased significantly: now $1 per million input tokens and $5 per million output tokens, quadruple the price of the previous Haiku.
  • It lacks the image analysis capabilities of the previous version, and Anthropic plans to keep both versions of the same series available for “users requiring maximum cost-efficiency and image processing.”

You cannot yet access Claude 3.5 Haiku on Claude web and app. It’s currently available for testing through Anthropic's API and some third-party providers, including AWS Bedrock and Google Cloud's Vertex AI.

Lead with AI - Icon

Available Now: Lead with AI PRO.

We’ve been amazed by the reception of Lead with AI, our course with over 170 senior leaders from companies like McKinsey, Apple, Gartner, Microsoft, and Toyota who graduated.
​
But AI never stands still, which is why we’re launching Lead with AI PRO.

Upgrade to Lead with AI PRO today and get:

Active Online Community

‍Discuss your AI implementation with peers in sub-communities dedicated to your focus area, from HR to Creative and Marketing.

Research and Video Library

From a Masterclass on NotebookLM to creating AI Marketing agents, access over 30 in-depth video lessons and exclusive AI research reports.

Monthly Masterclasses

Learn directly from industry experts in live interactive masterclasses about new tools, advanced strategies, and optimizing AI platforms like ChatGPT, Copilot, and leading specialized solutions.

Want to always be ahead – no matter how fast AI develops?

​Join today and get your second month FREE.

Upgrade to Lead with AI PRO now.
Lead with AI - Icon

How Wendy’s Nails Their Enterprise AI Experiments

Wendy’s is making headlines with its latest AI adoption through Palantir Technologies, optimizing supply chain management for its strategic $1 Frosty promotion. With Palantir's AI, Wendy’s can predict ingredient shortages and avoid costly overstock, enabling stores to keep up with high demand while maintaining profitability on slim margins.

But this isn’t their first successful AI implementation. Earlier, Wendy’s introduced FreshAI, powered by Google Cloud, which improved drive-thru service speed by an average of 22 seconds.

These two initiatives highlight three key lessons for those considering business AI deployment:

  • Target High-Impact Areas: In both examples, Wendy’s implemented AI to solve their "tough problem" ripe for improvement, channeling AI efforts where it could most significantly enhance the customer experience and crew efficiency in alignment with defined business goals.
  • Prioritize “End-User” Experience: In a recent interview with Fortune, Wendy’s VP of Restaurant Technology, Matt Spessard, shared they hyper-focused on improving areas for both their crew and the customers. This approach allowed them to tailor the AI application to create a differentiated customer experience while sticking closely with the team’s workflows and language.
  • Start with Pilot Programs: Wendy’s began both AI projects in select locations with long piloting periods, closely monitoring results and gathering feedback from employees and customers to adjust before expanding. This “test and refine” approach allowed them to fine-tune AI applications and ensure they work well in broader adoption.

T-Mobile Pays OpenAI $100M to Build AI Customer Service

T-Mobile's New AI Service May Soon Handle Support and Make Account Changes for You - CNET

American telecommunications company T-Mobile's $100M partnership with OpenAI on its IntentCX platform over the next three years represents a major investment in AI-driven customer service. T-Mobile plans to begin implementing IntentCX in 2025, following thorough testing and refinement.

Here’s what stands out:

  • AI-Driven Solution: T-Mobile is leveraging OpenAI's generative technology to create IntentCX beyond rule-based responses. This system analyzes customer intent and applies real-time decision-making to resolve issues autonomously.
  • Proactive Service Capabilities: Integrated into T-Mobile’s systems, IntentCX can preemptively address customer needs, even executing tasks automatically when permitted. This means customers won’t just get automated responses but will receive tailored actions.
  • Potential Expansion: T-Mobile hints at extending this AI solution to other industries, positioning IntentCX as a blueprint for AI-powered customer engagement on a broader scale.

With T-Mobile’s substantial investment, other major companies may follow suit, which could accelerate the shift toward automated customer service. This is one of the largest enterprise contracts of OpenAI, and IntentCX’s success could signal the phasing out of traditional customer service roles, especially as it proves capable of handling complex interactions.

​Is Trump Good for AI? Fast Gains & 'Black Swan' Risks for Business​

In this Newcomer podcast episode, Silicon Valley journalist Eric Newcomer and Cerebral Valley AI Summit co-hosts Max Child and James Wilsterman tackle the pressing question on many minds post-election week: what does Trump’s victory mean for AI?

With a wide-angle view of the topic, here are five key thoughts they covered in the first 28 minutes:

  • Energy Policy Support for AI: Trump’s administration could benefit AI development through support for energy initiatives, especially nuclear and solar power. AI requires immense energy, and a looser regulatory stance on energy could help fuel the power needs of large AI models and data centers.
  • Chip Manufacturing and Trade: Trump has a different view from the CHIPS Act, favoring tariffs over incentives to bring semiconductor production to the U.S. This could make AI hardware production more resilient but adds tension with China, especially concerning Taiwan, a critical source of chips for AI advancements.
  • Deregulation and Innovation: Many tech leaders anticipate that Trump might limit AI regulations, allowing companies more freedom to innovate. This would contrast with the Biden administration’s recent executive orders, which added oversight for large AI models. There’s a belief that Trump may adopt a "let AI develop without interference" stance, potentially accelerating advancements—but with safety risks.
  • Increased M&A Activity: Under Trump, the merger and acquisition environment could become more favorable, which might allow tech giants to acquire smaller AI firms and integrate their innovations. While this could drive rapid growth, it raises concerns about monopolization and reduced competition in AI tech.
  • “Black Swan” Risk of Instability: A recurring theme during the discussion was Trump’s "chaos factor." Trump’s unpredictable policies could create instability, particularly if foreign relations with China worsen or unexpected events lead to sudden policy shifts. Such instability could hinder AI progress, especially if global supply chains or regulatory landscapes are disrupted.
Lead with AI - Icon

Quick Hits

  • Another winner on the election night was Perplexity. Its Election Hub gained almost 4 million views and positive engagement from election watchers.
  • OpenAI bought the domain name “chat.com” for likely over $10 million from HubSpot co-founder Dharmesh Shad, paying him in… shares. That’s after the back-and-forth between OpenAI and Elon Musk’s AI company over another pricey domain, AI.com ($11 million).
  • Ai-Da Robot, the world’s first ultra-realistic robot artist, made history by selling a portrait of Alan Turing for $1.08 million at Sotheby’s, far exceeding the pre-sale estimate of $180,000.
  • ​Microsoft has filed a patent to beat AI hallucinations, proposing a new method called the "Response-Augmenting System" (RAS). RAS allows AI models to access and evaluate information from external sources and user feedback, flagging responses if they appear potentially inaccurate.
  • ​ChatGPT's rise has disrupted Chegg's business as students switch to free AI tools for homework help, leading to over half a million lost subscribers and a 99% drop in Chegg’s stock. Despite introducing its own AI, Chegg's struggles underscore the transformative impact of generative AI on traditional education models.
  • ​Salesforce plans to hire over 1,000 employees to boost sales of its new AI tool, Agentforce. CEO Marc Benioff highlighted strong initial momentum since Agentforce's recent launch, priced at $2 per agent conversation.
  • Coca-Cola’s iconic “Holidays Are Coming” ad is fully AI-created for the first time (Watch it here). Beyond ads, Coca-Cola has long integrated AI across business functions, including customer engagement, vending optimization, and consumer insights. Recently, it invested $1.1B in a five-year partnership with Microsoft for advanced cloud and generative AI capabilities.
  • ​The Washington Post launched “Ask the Post AI” that allows you to draw knowledge and insights from the publication’s archives.
Lead with AI - Icon

From The Lead with AI Community

Every day, Lead with AI members discuss practical ways to benefit from AI in their work and organizations.

This week's highlights include:

  • Miriam shares a hilarious example of when you send your AI clone to a meeting.​
  • Hendrick shares his new process of creating a presentation intro clip combining NotebookLM, ChatGPT, Gamma AI, Prome AI, and VideoLeap.
  • Helen shares her mini demo with Anthropic Computer Use Agent to collate 401K rollover information across bank websites, forms, and resources.
Join Us Here
Lead with AI - Icon

If you made it this far, reply and tell me what you'd love AI to take over in your daily workflow.

Also, please forward this newsletter to a colleague and ask them to subscribe.

If you have any other questions or feedback, just reply here or inbox me.

See you next week,

Daan van Rossum - Lead with AI

Daan van Rossum​

Host, Lead with AI

​Microsoft introduces new multi-agent system Magnetic-One​

The future of having your truly “AI team” is here!

Microsoft researchers just announced Magnetic-One, a generalist multi-agent system that can carry out complex, multi-step tasks by itself across the web and local files.

Magnetic-One uses a lead agent called Orchestrator agent, which is similar to Salesforces’ Atlas reasoning engine in Agentforce, plans how to conduct a task, directs the other four specialized agents, monitors progress, and adjusts plans as needed to make sure “the AI team” is completing a given task.

  • WebSurfer: Manages tasks on Chromium-based web browsers.
  • FileSurfer: Reads local files and navigates directories.
  • Coder: Writes code, analyzes information, and creates new artifacts.
  • Computer Terminal: Provides a console shell for executing code and installing programming libraries.

The system is available for testing within Microsoft’s AutoGen framework and can be used with different reasoning LLMs.

Besides that huge release, Microsoft also:

​Claude 3.5 Haiku Sparks Debate​

Anthropic released its new lightweight LLM Claude 3.5 Haiku, which is:

  • Surpassed its predecessor Claude 3 Haiku and even the large Claude 3 Opus on intelligence and safety benchmarks.
  • Performing faster and better on coding suggestions, data extraction and labeling, and content moderation.
  • Having longer text outputs and a more recent knowledge cutoff (July 2024).

But the upgraded version also raised some eyebrows as:

  • Despite being positioned as a "budget" model, the cost has increased significantly: now $1 per million input tokens and $5 per million output tokens, quadruple the price of the previous Haiku.
  • It lacks the image analysis capabilities of the previous version, and Anthropic plans to keep both versions of the same series available for “users requiring maximum cost-efficiency and image processing.”

You cannot yet access Claude 3.5 Haiku on Claude web and app. It’s currently available for testing through Anthropic's API and some third-party providers, including AWS Bedrock and Google Cloud's Vertex AI.