Reach 10k+ professionals leveraging AI. Sponsor this newsletter (available from May).

Hi there, and welcome to the 228 new subscribers who joined last week.

Curious about making custom GPTs? Here’s my best tips + some cool tools for GPT builders.

What’s brewing in AI #34

• Dario’s Picks: OpenAI adds image editing to DALL·E and leading AIs add ability to interact with external apps

• In Focus: From knowing to allocating

• GPTs: the latest top arrivals in the GPT store

• Bytes: the other AI news this week you don’t want to miss

 Dario’s Picks 

I. OpenAI adds image editing to DALL-E

OpenAI just added two new features to DALL·E:

1. Edit images directly inside DALL-E across web, iOS, and Android. You can select parts of the image to re-generate based on a new prompt.

2. Quick buttons for different styles (sci-fi, gothic, low angle, ++) and aspect ratios.

Why it matters

DALL-E, being part of ChatGPT, is one of the most used image generators right now. It’s finally catching up with some of the editing and convenience features of its more advanced competitors, including Midjourney.

II. Anthropic and Cohere allow external apps to interact with their LLMs

In a similar way that custom GPTs can be configured to use third-party APIs (allowing for things like scheduling meetings, sending emails, retrieving stuff from databases, etc.) leading LLMs are being upgraded to allow tool integrations:

• Claude is now capable of tool use: the new feature enable developers to give Claude access to external tools to perform tasks.

• Cohere announced their new model Command R+, an affordable LLM focused on business applications. It has advanced Retrieval Augmented Generation (RAG) features and also capable of tool use (connecting to external apps).

Why it matters

We can tell a lot about the future of AI apps by looking at the capabilities developers get to work with. Leading models being upgraded with Tool Use/function calling shifts the role of LLMs from just ingesting and outputting text, to actually being able to take actions in the real world. Developers can create multi-step workflows, allowing LLMs to use several tools to execute complex tasks – without human interference.

We’re moving from simple chatbots to fully-fledged AI agents.

This issue is brought to you by

Hire a world class AI team for 80% less

Trusted by leading startups and Fortune 500 companies.

Building an AI product is hard. Engineers who understand AI are expensive and hard to find. And there's no way of telling who's legit and who's not.

That's why companies around the world trust AE Studio. We help you craft and implement the optimal AI solution for your business with our team of world class AI experts from Harvard, Princeton, and Stanford.

Our development, design, and data science teams work closely with founders and executives to create custom software and AI solutions that get the job done for a fraction of the cost.

P.S. Wondering how OpenAI DevDay impacts your business? Let’s talk!

 In Focus 

From knowing to allocating

Last week I read a great article by Dan Shipper called The Knowledge Economy Is Over. Welcome to the Allocation Economy.

A key skill of pretty much any office worker for the last decades has been what you know – and the ability to recall, share and apply that knowledge. But AI is quickly making this skill redundant as it’s better and quicker at it than you and me. Managing resources/models is the new most-valuable skill.

We’re moving to what the author calls the allocation economy, where being able to manage, prioritise and allocate resources will be more important than knowledge. Specifically, the most valuable skill for everyone using AI in their job will be as managers of models. Today’s managers already practice this skill every day, just with people and not models.

The author argues that we, the “model managers”, will need these skills:

• a coherent vision: articulate, concise and rooted in a clear purpose.

• a clear sense of taste: knowing what good results should look like.

• ability to evaluate talent: judging which models are best at which tasks.

• knowing when to get into the details: not micromanaging, not over-delegating.

My take on this

Fully agree with the skills mentioned becoming more important, and knowledge becoming less valuable.

Can we call it model leadership, though, rather than model management? If we’re really headed towards AGI, inspiring and guiding AI seems a step above merely controlling it. Because who actually likes being managed? ;)

 GPTs 

Top new arrivals in the GPT store

Highly rated, new GPTs featured in OpenAI’s official GPT store (from the last week)

• Academic Writting #16 in  Writing

• Video Summarizer #13 in  Research

Top newcomers on whatplugin.ai

Highly-rated GPTs that made it into whatplugin.ai from the last week. How rankings work.

Submit your GPT to whatplugin and get featured in the newsletter.

 Bytes 

• 🔥 Fresh off the whatplugin blog: Review of the best AI language tutor apps (by Denise)

• 10 ways of getting started in AI – without being technical.

• ChatGPT Enterprise customers has 4x’ed since January. ndicative of the speed at which big organisations are embracing AI.

• Google might start charging for AI-powered search soon.

• Business Schools are going all in on AI.

• Billie Eilish and 200+ artist sign an open letter from the Artist Rights Alliance calling for AI developers to respects artists’ work.

Research

• Huge market map of machine learning, AI and data companies released by VC firm Firstmark. You can also read the full post.

• Survey about which AI’s people use for work, by FlexOS. The top 100 gen AI tools grew on average 9% in web visits over the last two months. ChatGPT, Google Gemini, and Microsoft Copilot were the most used. Image generators are also popular at work, particularly Canva’s AI suite, Leonardo AI and Midjourney.

• S&P AI Benchmark: a benchmark ranking LLMs by their ability to understand and solve business and finance problem. GPT-4 Turbo is at the top of the list.

Product

• OpenAI introduced new features to make it easier to fine-tune using the GPT-3.5 API. This enables developers to tune the model to understand content more deeply and improve performance on specific tasks.

• Stability has released Stable Audio 2.0. It can produce high-quality, full tracks with coherent musical structure up to three minutes.

Partnerships

• Yahoo acquires Artifact, an AI tool for news recommendations by the co-founders of Instagram.

• AWS + Mistral AI partnership to bring Mistral Large on Amazon Bedrock.

 The wizard’s favourite AI newsletter 

what i’m reading right now

simple.ai 🕵🏻‍♂️ - deep-dives on AI by Hubspot’s co-founder (highly recommended)

The Neuron 😸 - easy weekday read on AI’s latest developments

Bagel Bots 🥯 - best hands-on tips & tricks

AI Minds 🧠 - semi-technical breakdown of trending AI topics

*if you sign up for this newsletter I may earn a commission at no cost to you. If you do, THANK YOU, it will make it possible for me to continue to do this.

 Super short survey