The TLDR

Highlight: ChatGPT just dropped GPT Agents! Agents solve one of ChatGPT’s biggest problems by allowing continuous contextualized environments. Read more below.

Musing of the Week: We love keeping tabs on Hollywood, and we revisit the SAG-AFTRA negotiations to consider the implications of their new tentative contract.

🛠️ Agents Fix ChatGPT’s Biggest Problem

Last week we talked briefly about the upcoming release of GPT Agents, and this week the tools dropped for Plus and Enterprise users. This new rollout presents a tremendous upgrade in managing repetitive automate-able tasks. To demonstrate some of the basic utility, we revisited an old project:

Supercharging Customer Insights Revisited

Months ago, we demonstrated how ChatGPT could be used to generate customer intelligence based on user reviews. While we got some great analysis, we had to hand-crank through some elements. One of the biggest barriers to really using this among a team was that we had to remake the environment for every new work session. This meant that we could only share final insights and wouldn’t be able to give someone the ability to ask questions later.

Creating an agent solved this problem

With a ChatGPT agent, we were able to quickly set up an ongoing environment that could answer any questions we had based on the data without the constant need to reset and re-upload. OpenAI gives us two ways to set this up:

1. Through Conversation:

   • Start by talking to Chat about your goals and how you want it to help you.

   • This approach is intuitive, and as the agent learns and adapts, it refines its understanding and asks you for the context it needs to achieve your goals.

   • As always, Chat makes some incorrect assumptions, but it’s a great way to get started.

2. Direct Model Configuration:

   • Configuration offers a way to precisely define the agent's parameters, capabilities, and context.

   • This involves creating specific prompts and uploading your context, you can even add actions that allow Chat to reach information off the platform.

By allowing us to create agents, OpenAI has addressed a crucial limitation of their product: the need for repetitive setup and contextualization. 

Regardless of the route you take to set them up, these agents are easy to build and provide a more dynamic, user-friendly approach to leveraging AI for complex tasks. The impending launch of the agent marketplace will open up a huge range of user-generated GPT Agents that could provide an out of the box solution for many tasks, at no additional cost (at time of writing). More to come! We’ll keep messing with GPT Agents in the upcoming weeks.

🧠 Musing of the Week

Last week, SAG-AFTRA finally reached a tentative agreement with the AMPTP. We've talked about the AI-related implications of the ongoing negotiations in previous issues, and the most recent SAG-AFTRA contract seems to be a step toward providing actors with some protections against the unauthorized use of their images while also offering new potential for them to make income from licensing.

However, not everyone is happy with the new deal. On the surface, the contract seems great for well-known actors. But it also likely means less room for new actors, as established stars could appear in multiple projects at once using their digital replicas. Then, there's the concept of "Synthetic Performers" – actors created entirely by computers and owned by the production companies.

This raises a question: Is digital reproduction of human performers really the biggest concern for the industry?

Considering how young audiences are already growing accustomed to virtual influencers, it may be the “Synthetic Performers” who pose the greatest threat. The success of actors as prominent public figures has always been, at least in part, an artifact of the enormous force of the entertainment industry backing them. In a new context where audiences are constantly demanding more content, faster turnaround times, and reduced costs, these digital performers may be cheaper, simpler, and more efficient from the studios’ perspective. If we see the industry turning its engine away from human performers and towards the creation of the “Synthetic Performers,” it would represent a fundamental change in industry dynamics, not abated by these new deals.

🙌 If you’re hyped about the generative AI industry specifically, here are some of the coolest roles we’ve seen this week:

• AI Strategy Expert @ SAP

• Senior AI Analyst @ Florida Blue

• AI/ML Lead, Go to Market, US Cities @ Amazon Web Services

• Community Marketing Lead @ Kena AI

🔨 Check out these other AI tools we’ve been looking at this week (creator edition):

• SmiliMedia: Turn longer videos into optimized, captioned Reels and Shorts

• StockMusic: Royalty-free AI music for creators

• Vidiofy AI: Turn articles into short videos and expand your reach

That’s all for this week. See you next Tuesday!

Lorel & Reily