Microsoft wants you to build your own AI co-worker; will you?

Welcome to today’s edition of The Tensor, get smarter with executive insights on the latest in AI & Tech Industry, 5 min reads, 3x a week.

It’s AI Agents all around in today's Tensor:

• Top Story - Microsoft wants you to build your own AI co-worker with Copilot Studio

• Industry - Salesforce announces Agentforce to build your AI sales team.

• Research - Small but Mighty: How Tiny AI Models Are Making a Big Difference

• TLDR - Runway API for Video Generation AI, New AR glasses from Snapchat and 2 more.

Let’s dive in

Microsoft wants you to build your own AI co-worker with Copilot Studio, Will you?

The scoop: Microsoft introduced "Copilot" a few years back, weaving its assistance into tools like GitHub and Microsoft 365. Now, they're jumping on the AI Agent train with Copilot Studio—the latest platform that lets businesses craft custom AI agents tailored to their needs. In an industry buzzing about AI agents, Microsoft's saying, "Why not build your own?"

How it works:

• No-Code AI creation platform: Copilot Studio enables anyone to create AI assistants that automate routine tasks and streamline workflows—no coding required. Making internal tools becomes a lot easier when all you have to do is use English.

• Right Where You Work: Microsoft is leveraging its biggest advantage, being your default at work. It allows your AI agents directly into Microsoft 365 apps like Teams and SharePoint, so they assist you within the tools you already use daily.

• Real-World Impact: Imagine an AI agent that schedules meetings, summarizes lengthy reports, or even drafts emails based on your guidelines—all designed by you for your team's specific needs, for each team.

• Azure Inside: Built on Microsoft's robust Azure cloud platform, these AI agents come with enterprise-grade security, scalability, and compliance baked in, allowing for easier adoption with enterprises.

Why it matters: Microsoft has been all-in on AI, aiming to make it as ubiquitous in business as Word and Excel. With over 50,000 organizations already adopting Copilot Studio, the reach is substantial—and growing. This move could redefine workplace efficiency, giving companies the tools to automate like never before.

Bottom line: Microsoft's Copilot Studio is its latest attempt to introduce an AI assistant into today's office reality. The question I have is, in a world where people fear that AI is going to take their jobs, will they adopt a system which creates the AI that automates their job?

Salesforce announces Agentforce to build your AI sales team 

The scoop: Salesforce, the titan behind the world’s leading CRM platform, is also jumping on the AI Agent bandwagon with the launch of Agentforce. Agentforce aims to automate tasks like managing customer inquiries, qualifying sales leads, and optimizing marketing campaigns. Whether you're in sales, customer service, or marketing, this AI tool aims to streamline your daily operations and boost productivity.

Highlights:

• AI Agents That Take Action: Unlike standard chatbots that merely respond to queries, Agentforce’s AI agents actively execute tasks within your CRM. From updating customer records to initiating follow-up emails, these agents handle the grunt work.

• Low-Code, High Efficiency: No need to hire a squad of developers. With user-friendly, low-code tools, Salesforce wants your sales team to build and customize AI agents tailored to your specific needs.

• Deep CRM Integration: The biggest strategic play in my opinion is that Agentforce plugs directly into Salesforce's rich data ecosystem, giving AI agents context and insight that generic solutions can't match.

• Powerful Partnerships: Salesforce is built on partnerships and they are leveraging it with tech giants like AWS, Google, and IBM, to extends capabilities and bring in a suite of pre-built actions to enhance your workflows.

Why it matters: Sales is all about pipeline creation and yield. Agentforce wants to make it easier to do so. If Salesforce can pull it off, it wants to be at the center of a world where your sales reps wake up to qualified leads, your customer service resolve issues before they escalate, and your marketing campaigns optimize themselves in real-time. It's like adding an all-star player to your team who never sleeps and always knows the playbook. I am looking forward to seeing it play out.

Small but Mighty: How Tiny AI Models Are Making a Big Difference

The scoop: You might think that giants like ChatGPT and Claude run the show in AI, but small language models (SLMs) like BERT and DistilBERT are quietly powering many of the apps and services you use daily. From voice assistants on your smartphone to recommendation engines on streaming platforms, these lean models are the unsung heroes making AI accessible and efficient. A recent paper helped understand why.

Highlights:

• Team Players in Disguise: Models like MobileBERT and TinyBERT excel in tasks like text classification and sentiment analysis, often outperforming larger models in speed without sacrificing much accuracy. For example, your email's spam filter likely relies on these efficient models to keep your inbox clean.

• Efficiency Experts: Small models shine on smartphones and IoT devices. They enable quick language processing without heavy battery use or needing constant internet. For example, SqueezeBERT uses very little computing power, making it great for smart home devices like Alexa and Google home that quickly respond to voice commands.

• Specialists in the Field: In healthcare, models like ClinicalBERT and MedPalM 2 are tailored for medical jargon, providing more accurate diagnostic support than general-purpose LLMs. Their interpretability also makes them more trustworthy in sensitive applications like finance and law.

• Perfect for Tight Spaces: Edge devices benefit from SLMs that can run complex AI tasks locally. Think of LiteXLNet powering language understanding on your smartwatch, enabling features like voice-to-text without sending data to the cloud.

Why it matters: Edge computing is set to explode, with Apple, Google, Samsung and others shipping millions of devices with efficient AI models that can run locally. Small models are not just cost-effective alternatives; they're essential for bringing AI capabilities to smartphones, wearables, and other edge devices. This decentralization creates an amazing experience for the user with complete privacy.

Bottom line: In the world of AI, bigger isn't always better. Just as a pocket knife can be more practical than a chainsaw for everyday tasks, small models are proving they can deliver big results where it counts the most—right in the palm of your hand.

• Runway, one of several AI startups developing video-generating tech, today announced an API to allow devs and organizations to build the company’s generative AI models into third-party platforms, apps, and services.

• Slack announced new AI-powered features, including AI agents and improved integrations with Salesforce.

• Groq announced a partnership with Saudi Aramco to build the world’s biggest AI inference cluster

• Google shipped a series of changes which have significantly improved the Gemini 1.5 Flash latency (>3x reduction) and output tokens per second (>2x more)

• Snapchat released new AR spectacles aimed at developers for $99 a month offering ground breaking AR technology.

Thank you for reading today’s edition of the Tensor!
Someone forwarded you today’s edition and you want to read more?