🧙🏼 Microsoft's autonomous agents

Howdy, wizards.

While the lawsuits keep coming, Perplexity keeps shipping. They’re launching reasoning powered search in beta, where the model will spend several minutes browsing and researching when it is prompted with complex tasks. I expect OpenAI is going to implement a similar auto-switching of models eventually too.

Here’s what’s brewing in AI today.

DARIO’S PICKS

1. Microsoft’s autonomous agents in Copilot Studio

Starting next month, Microsoft is letting users create custom AI workers in Copilot Studio, as the agent builder moves from private to public preview. Based on OpenAI’s o1 models, agents can be used to automate tasks of various levels of complexity – from simple prompt-and-response to multi-step business processes.

Microsoft is also launching 10 pre-built agents inside Dynamics 365, including agents for sales, service, finance, supply chain and more.

Businesses will be able to set up their own guardrails and controls for each agent within Copilot Studio as well, such as data loss prevention, authentication protocols, and more.

Here’s some companies that have been testing the private preview:

• Pets at Home: Developed an agent for profit protection, potentially saving seven figures annually.

• McKinsey & Company: Created an agent reducing client onboarding lead time by 90% and administrative work by 30%.

• Thomson Reuters: Built an agent accelerating legal due diligence, completing tasks in half the time.

‎ Why it matters‎ ‎ Microsoft’s calls these agents “the new apps for an AI-powered world” which indicates they’re highly confident in their impact. If you’re trying to wrap your head around what these agents really are, you can think of custom GPTs for ChatGPT, but with better reasoning capabilities, totally integrated with Microsoft’s ecosystem, and with secure access to your business data.

TOGETHER WITH HUBSPOT

DARIO’S PICKS

2. Four real-world use cases of AI in media & entertainment

I’ve taken a broad look at the media & entertainment industry to find some interesting ways companies are applying foundation models.

Here’s some interesting use cases from recent months:

• VG, a leading news outlet in Norway, developed an internal AI toolbox with about 25 generative AI tools. These tools enhance journalists' ability to write, research, and investigate more efficiently. For example, they have a tool that extracts information from text and visualizes relationships between entities as network graphs, which helps journalists in understanding complex connections. Another tool, allows journalists to query Statistics Norway's data, automatically retrieving relevant datasets and generating visualizations using Datawrapper. These tools help journalists dive deeper into stories and present data in more engaging ways. SOURCE (JUNE 2024)

• Jumpcut Media, a company making AI-powered tools for the entertainment industry, uses Claude 3 Opus to power its ScriptSense product, which generates comprehensive script coverage—a specific report for a screenplay including scene summaries and character breakdowns. This helps Hollywood studios, agencies, and production companies save time by evaluating scripts in seconds. ScriptSense also indexes scripts in a searchable database, aiding creative teams in meeting studios’ requirements for which content they need. SOURCE (JULY 2024)

• Scribd, a digital document library, uses Claude from Anthropic to automatically generate high-quality metadata for over 100 million user-uploaded documents across its platforms Scribd and SlideShare. This improves content discoverability and drives user engagement, leading to a 7% increase in subscriptions. The AI-generated metadata also enhances SOURCE (SEPTEMBER 2024)

• Full Fact, a UK-based nonprofit working in 18 countries to combat misinformation, is using Google’s gen AI solutions to actively monitor stories, allowing its 30 fact-checking partner organizations to focus on handling specific claims and harmful information. SOURCE (SEPTEMBER 2024)

‎ Why it matters‎ ‎ The ways in which media companies are using AI is diverse. Unsurprisingly, it’s being used to automate routine tasks like summarising, indexing and flagging content for different purposes. It’s also helping to create better content, with creative applications like the ones from VG – using AI to make complexity easier to understand both for their writers and readers.

FROM OUR PARTNERS

Escaping AI POC purgatory: Techniques for enterprise AI engineers

Many companies struggle to move generative AI from experimentation to production.

Join us Oct. 29 at 9am PT. Sam Julien, Writer's Director of Developer Relations, will share practical strategies to overcome enterprise AI engineering challenges using a full-stack approach.

Topics include:

• Managing project scope

• Improving accuracy with domain-specific models & graph-based RAG

• Navigating AI application development

Can’t make it live? Register anyway and we’ll send you the recording.

Register now

THAT’S ALL FOLKS!

Affiliate disclosure: To cover the cost of my email software and the time I spend writing this newsletter, I sometimes link to products and other newsletters. Please assume these are affiliate links. If you choose to subscribe to a newsletter or buy a product through any of my links then THANK YOU – it will make it possible for me to continue to do this.