The Quest to Rename AI

Welcome back to the Enterprise AI Playbook, Issue 3. Here are the successes, cautionary tales and deep dives from this week.

Successful launches - Apple reveals its Gen AI plans with Apple Intelligence

While its early to tell if this was a successful launch, there were a number of interesting aspects revealed in the WWDC keynote yesterday.

• Text editing across applications, like email, notes and third party apps

• Image generation across multiple apps and three different default styles

• More complex question synthesis and response generation, named actions, with 100s of built ins

• Semantic index used for RAG applications, accessible across apps

• On device and cloud based LLMs, with a promise of routing

• Private cloud compute for running Apple Cloud LLMs on apple silicon

• Improved Siri through context, in-app extraction capabilities and typing

• Image editing and search within photos and other apps

• Integrating ChatGPT and other LLMs into the platform

Key takeaways:

• Apple is building expectations at the OS level for how to use Gen AI. This will put pressure on other OSs and applications with new expectations for search and edit functionality from mainstream users

• Return of intents and built in actions. Intents are coming back to the mainstream as a way to handle specific use cases and desired actions.

• Tensions exists between privacy first and ecosystem. With the addition of ChatGPT-4o and the promise of other models in the future (like Gemini), it’s unclear who the bad guys who take your data are.

• Siri knowing all - with the semantic index release that’s accessible to Siri, data access will become a larger question. One example Apple provided was copying information from a photo of a driver’s licence. This index and user pattern makes it easier for someone to access private information and also changes user patterns for storing and accessing sensitive information. Will be interesting to see if this option is default off for corporate accounts, along with the other Gen AI features.
  

Link to the full Keynote

Cautionary Tales - Magnitude 4.2 for creators

This past week Adobe has been under fire for their Section 4.2 for the creative suite.

4.2 Licenses to Your Content. Solely for the purposes of operating or improving the Services and Software, you grant us a non-exclusive, worldwide, royalty-free sublicensable, license, to use, reproduce, publicly display, distribute, modify, create derivative works based on, publicly perform, and translate the Content.

While it’s up to the legal system to determine what falls under this agreement, users on social media with one post on X/Twitter, generating almost 20M views. Adobe has issued a clarification post, but it is unclear how impactful it will be on customer retention and rebuilding the brand. Slack and Zoom had similar experiences last year and this year respectively.

Deep dive - 15 Customer Experience Metrics for AI

The success section was quite a deep dive, but I also sat down with the Head of AI support at Voiceflow, Tahsim, to discuss important metrics for building customer support automation. In this hour long video we cover:

1. Which metrics to prioritize for your stage (15 total)

2. Analyzing metrics from AI Agent Tico

3. Which teams are responsible for each metric

Link to the full video

Questions to ask your team

How does your service stand out as Gen AI moves to the operating system level?

Until next week, Denys - Enterprise AI @ Voiceflow