OpenAI Restructuring, $8.2B for AI Agent Startups, and New Tesla Products

PLUS HOT AI Tools & Tutorials

Oct 11, 2024

Hello and welcome to our weekly roundup!

What do you think of the Tesla event? It's kind of fun, and it feels like Elon is taking concepts from Cyberpunk and bringing them into real life. And one would hope we're limited to futuristic cars and helpful robots without all the other anti-utopian stuff.

In addition to Tesla, we have accumulated a whole bunch of news about the future of OpenAI this week. There are both good points and those that cause some anxiety.

Let's go through them all.

The other day, we also experimented with ChatGPT Canvas and shared some of the top tips and tricks that might come in handy for you:
Canvas Review: New Level For ChatGPT?
Stepan Ikaev
·
Oct 10
Read full story

This Creators’ AI Edition:

Featured Materials 🎟️
News of the week 🌍
Useful tools ⚒️
Weekly Guides 📕
AI Meme of the Week 🤡
AI Tweet of the Week 🐦
(Bonus) Materials 🎁

Featured Material 🎟️

OpenAI's Plans and Goals for the Near Future

This week, we have several topics related to the future of OpenAI. And I should warn you that some of the information is based on anonymous sources (albeit connected to reputable media outlets), so please don't take this as the absolute truth.

According to the Financial Times, OpenAI plans to abandon its non-profit roots and restructure itself into a “public benefit corporation.” Insiders say this decision will allow the company to shield Sam Altman and his supporters from unwanted takeovers and objections to how they run the company. Under the new structure, OpenAI leaders will weigh both public impact and profit in making decisions.

Two other AI companies operate under a similar system: xAI and Anthropic.

If the FT's information is confirmed, the nonprofit organization that has overseen OpenAI since its founding in 2015 will continue to exist but no longer manage its operations. After that, a situation like Sam Altman's abrupt firing last fall would become impossible.

Do you think OpenAI will change after Altman gets more serious powers? And, if it does, in which direction? Maybe Musk was right to start Game of Thrones inside OpenAI? So many questions and very few answers.
By the way, here are the details on what Ilon thinks about the mass exodus of OpenAI's top managers:
OpenAI’s Game of Thrones, ChatGPT Gets Voice Mode, and Meta Releases Llama 3.2
Stepan Ikaev
·
Sep 27
Read full story

Keep your mailbox updated with key knowledge & news from the AI industry

OpenAI Will Not Become Profitable Until 2029

Another insider comes from The Information. Citing OpenAI's financial documents, the journalists reported that the startup could suffer annual losses of up to $14B in 2026. The report said that revenue projections showed OpenAI would earn $100 billion in 2029 and spend about $200B over the rest of the decade, excluding stock compensation. The report also shows losses of $44B between 2023 and 2028.

Coupled with the previous news, it seems that OpenAI may be getting off on a shaky footing. With a substantial commitment to investors and responsibility for the entire company, Sam Altman must think carefully about every next move. After all, mistakes will be very costly. The good news is that OpenAI is quite sober about its prospects, which is encouraging.

OpenAI Opens Five New Offices Around the World

And finally, quite official news from the company itself. OpenAI announced in X that it is opening new offices in NYC, Seattle, Paris, Brussels, and Singapore in addition to its existing offices in San Francisco, London, Dublin, and Tokyo. In addition, the company also revealed that managing director Oliver Jay (former chief revenue officer at Asana and head of Asia Pacific and Latin America at Dropbox) will oversee international operations and drive global expansion from Singapore.

And this move looks quite expected. After a $6.6 funding round, OpenAI now has the opportunity to bring in global talent to its side. If you've been wanting to join the ChatGPT development team for a long time, now is the time!

News Of The Week 🌍

Tesla Held “We, Robot” Event

Elon Musk's company held an unexpectedly important “We, Robot” event in California. Tesla showed two electric car concepts—Cybercab and Robovan—and a new version of the Optimus robot. Cybercab is a steering wheel- and pedal-less autonomous cab with a futuristic design expected to cost less than $30,000. Production could begin by 2026, and the company is looking to launch a complete autopilot in Texas and California next year.

And Robovan is a van (wow!) with an equally unusual appearance that should seat 20 people. Production launch dates and costs for the second vehicle are still unknown.

As for Optimus, the company brought out a whole group of humanoid robots after the Robovan presentation. Musk stated that these robots can “do anything,” including cleaning around the house, walking the dog, or mowing the lawn. The estimated cost of Optimus is between $20,000 and $30,000 (but that could change). After the presentation, live footage showed people interacting with Optimus robots at tables and in the crowd.

The event turned out to be intense, so I recommend watching the footage:

https://www.youtube.com/live/6v6dbxPlsXs?si=jKvhrPgf5umh3htH

Writer Unveils Its First LLM Called Palmyra X 004

Writer, a San Francisco-based startup, has recently released Palmyra X 004, their LLM designed for enterprise use. This new model introduces several advanced features and capabilities that set it apart in the competitive AI landscape. Initial tests have shown that Palmyra X 004 performs well in function calls, also known as tool call. This allows the model to interact with external systems, execute complex workflows, and perform actions beyond simple text generation.

It already achieved 78.76% accuracy on Berkeley's Tool Calling Leaderboard, surpassing offerings from major tech companies by nearly 20%, and ranked in the top 10 on Stanford University's HELM benchmarks: 86.1% on HELM Lite and 81.3% on HELM MMLU.

Technical Specs:

128,000 token context window, matching OpenAI's GPT-41
Multilingual support for over 30 languages
Multimodal capabilities (beta) for processing text, images, and audio3

You can find more details about this model here.

Startups Developing AI Agents Raised $8.2B This Year

According to PitchBook, the number of deals among AI agent startups is up 81.4% year-over-year. Despite the impressive number, these companies feel good about this market. Analysts estimated that such AI startups raised $8.2B in the last 12 months. The most notable projects were Sierra, which developed customer service agents; Harvey, which offered AI agents for lawyers; and TaxGPT, an AI agent for processing taxes.

AI agents are becoming a mass phenomenon, at least in investors' portfolios. But whether the models will be able to fully replace lawyers, customer support, accountants and other roles, we will see in the coming months.

Sharing is caring! Refer someone who recently started a learning journey in AI. Make them more productive and earn rewards!

Refer a friend

Anthropic Launches API Message Batches

Following OpenAI, Anthropic has unveiled its solution for cost-effectively using AI models. The Message Batches API platform allows submitting up to 10,000 queries in a single batch for asynchronous processing. According to the company, the API is designed for tasks that do not require an immediate response, offering significant cost savings and improved scalability for enterprises working with large datasets.

Here are its key features:

Batch Size: Supports up to 10,000 messages per batch or 32 MB in size.
Processing Time: Responses are typically available within 24 hours.
Cost Efficiency: Offers a 50% discount on input and output tokens compared to standard API calls
Supported Models: Currently works with Claude 3.5 Sonnet, Claude 3 Opus, and Claude 3 Haiku

Pricing

Meta Releases Movie Gen for Video Generation

Meta recently announced Movie Gen, a new AI model for creating and editing video and audio content. The platform is called Movie Gen and offers four main capabilities: generating videos up to 16 seconds long from text prompts, creating clips featuring uploaded images (yes, you can turn your photo into a video), spot changes for existing videos, and overlaying sound effects to accompany videos.

What I particularly like about Meta's latest products is that they don't just launch interesting gimmicks but immediately offer real-world use cases. This makes Movie Gen look like a desirable solution.

SearchGPT Is Coming to ChatGPT This Year

According to Varun Shetty, OpenAI is bringing its AI search engine, SearchGPT into ChatGPT later this year. The update was confirmed at a media publisher event in Brussels, reported on by the Press Gazette. Shetty also revealed that the interface would include links back to publisher websites and accurate sourcing in addition to the natural language responses. With the open release of this product, OpenAI will become a full-fledged competitor to Google.

Great news, but given the presence of Perplexity, it doesn't seem like such a big deal. However, the hype around OpenAI will give the company a perfect start.
But will it be enough to beat Google?

Uber Plans to Launch AI-Powered Assistant for Drivers

Uber is preparing to launch an AI-based assistant developed in collaboration with OpenAI to accelerate the adoption of electric vehicles (EVs) among its drivers. The tool is powered by GPT-4o to process and respond to textual, audio, and visual inputs. The AI assistant will be available on the home screen of Uber's driver app, which is designed to provide personalized support to users transitioning to electric vehicles.

Initially, the assistant will focus solely on electric vehicle-related queries, but Uber plans to expand its capabilities to cover a broader range of topics by early 2025.

Another real-world scenario for applying AI to our lives. Sounds great, but I think it's too early to conclude. As always, I suggest waiting a bit and seeing the results.

Useful Tools ⚒️

Echo – Capture easier, think better with AI voice and text notes

Handinger – Extract data from the internet

JobJump – Master video interviews with live AI-tailored answer hints

Height 2.0 – The autonomous project management tool

Cove – Visual workspace for thinking with AI

Cove provides a visual, infinite interface for AI-assisted brainstorming and planning. Unlike linear chatbots, Cove allows for simultaneous idea generation and collaboration, mimicking the dynamics of real group work. The platform liberates well from the limitations of chatbots and document editors and allows content to be placed in a visual workspace for more intuitive interaction. This supports several AI models, including OpenAI, Meta, and Perplexity.