Copilot Wave 2: Bridging the Gap Between Humans and AI

Copilot Wave 2 is here, being dubbed as “UI for AI.” While the release notes provide a deep dive into all the technical details, the overarching theme is clear: Microsoft is moving to close the gap between humans and AI across various solutions, making it easier for teams to work together with their AI counterparts.

First and perhaps most interesting is the new Copilot Pages feature, which moves working with Copilot out of the ‘individual productivity’ category and into the ‘collaborative productivity’ category. This feature allows groups to work together on the same AI prompts and while the concept seems promising, the real-world application will ultimately determine how smooth this process is. Nevertheless, it can’t help but improve results over time as more humans learn from other humans how to talk to their AI co-workers.

As far as adoption goes, I also think it will help humans see AI as a co-collaborator rather than a novelty or a competitor. Jared Spataro called it a “human-to-AI-to-human” interaction.

Another important update comes with new features in Excel, allowing humans to see Copilot’s decision-making process when being asked natural language questions.

Similarly, Copilot for Python – sure to be a favorite of Data Scientists – provides transparency by showing its plan and the code it intends to run before running it.

Finally, Copilot Agents is a streamlined tool that uses Copilot Studio to help businesses pull relevant data from documents and take meaningful action based on that information. With RAG (Retrieval-Augmented Generation) capabilities, teams can leverage their existing work data more effectively and then use various connectors to act on that data.

One of the most notable quotes from Microsoft‘s announcement this week, which might have been missed among all the bells and whistles was: “Copilot is the best feedback loop for AI at work”. Given the number of Copilot for M365 users around the world right now, Microsoft is probably in the best position of any other company, based on feedback and anonymized telemetry, to know what is working and what is not.

If what we at Withum are hearing from our clients is any indication, these updates are going to be game-changers for AI adoption. The biggest challenge holding most organizations back has been moving beyond proof-of-concept (POC) projects to achieve real return on investment (ROI), but these new features could help make that leap.

As we continue to evolve our Copilot and AI practice at Withum, these updates only strengthen our optimism about the future.