OpenAI announces the enterprise version of ChatGPT

🥽 3 TRENDS

Meta announces Code Llama a rival to Github Copilot (🔗 link)

Meta has introduced Code Llama, a specialized language model for coding tasks that is built upon the Llama 2 model and trained to rival its main competitor Github’s Copilot.

The model supports various programming languages and offers features like code completion and debugging.

Available in three sizes—7 billion, 13 billion, and 34 billion parameters—the smaller models are optimized for low latency tasks, while the 34B model offers the best performance at the cost of higher computational power.

Big news from OpenAI! - Enterprise ChatGPT is finally here (🔗 link)

OpenAI has announced the launch of ChatGPT Enterprise, an enhanced version of its popular AI chatbot, designed specifically for business applications with features such as enhanced privacy and data analysis capabilities.

The enterprise edition competes directly with Microsoft's Bing Chat Enterprise and offers additional admin tools, including single sign-on, domain verification, and a dashboard for usage statistics, as well as customizable workflow templates.

ChatGPT Enterprise also includes unlimited access to Advanced Data Analysis, a feature that enables the chatbot to analyze data and generate insights from uploaded files.

OpenAI now allows developers to fine tune GPT3.5 (🔗 link)

OpenAI has introduced the ability to fine-tune GPT-3.5 Turbo for customized applications, with support for GPT-4 coming later this year.

Developers can now adapt the language model to specialized tasks, such as matching a brand's voice or formatting API responses in JSON.

The fine-tuning feature not only improves output consistency and instruction-following but also allows for up to 90% shorter prompts, speeding up API calls and reducing computational costs.

🛠️ 2 TOOLS

 FOR THE CONSULTANT

McKinsey launches new internal AI agent called Lilli. (🔗 link)

McKinsey and Company has launched Lilli, an AI-powered chat application designed to assist its employees by providing information, insights, and expert recommendations based on over 100,000 documents and interviews.

Developed by McKinsey's "ClienTech" team under CTO Jacky Wright, the tool aims to encapsulate the firm's collective knowledge to aid in consulting projects.

Named after Lillian Dombrowski, McKinsey's first female professional hire in 1945, Lilli has been in beta since June 2023 and is set to roll out company-wide this fall.

This is just the beginning of business specific AI agents built and trained on proprietary LLMs from open source models..

 FOR THE AUDIOFILE

Eleven Labs text to voice AI tool launces out of Beta (🔗 link)

ElevenLabs, an AI platform specializing in synthetic voice creation, has officially launched with support for over 30 languages, using an in-house developed AI model that can identify languages and generate emotionally rich speech.

The platform also offers a voice-cloning tool, allowing users to generate speech in multiple languages without typing text, thereby advancing its mission to make content universally accessible.

While the platform gained quick popularity for the quality of its voices, it has also faced controversy for being exploited by bad actors to spread hate speech and mimic celebrities.

💡 1 TIP

The Potato Prompt - a tip to push chatbots to get to the point.

The Potato Prompt, or "protato," aims to make AI responses concise and direct.

It cuts down on the verbosity commonly found in AI interactions.

For example, asking "Who is Joe Biden?" without the potato prompt would yield a long rambling answer.

With the potato prompt in place you get a brief answer like "46th President of the United States. Democrat. In office since 2021."

This tool streamlines communication with AI models and can be really handy when trying to get answers quickly.

To go back to the normal verbose AI assistant simply type in “unpotato”

Below is the original prompt you can add to ChatGPT 3.5 or 4 before you ask any questions. Give it a try and let us know what you think.

##

“Starting now you are very succinct. You don’t apologize if you don’t explain stuff right, and you don’t add additional info to eventually get to the point. You use straightforward language. You say the absolute minimum to get to the point. If you understand, say “got it - i am potato”.

But when I say "unpotato" then please revert to ignore this command and be a normal AI assistant as usual. “

##