10x Your Outbound With Our AI BDR

Imagine your calendar filling with qualified sales meetings, on autopilot. That's Ava's job. She's an AI BDR who automates your entire outbound demand generation.

Ava operates within the Artisan platform, which consolidates every tool you need for outbound:

• 300M+ High-Quality B2B Prospects

• Automated Lead Enrichment With 10+ Data Sources Included

• Full Email Deliverability Management

• Personalization Waterfall using LinkedIn, Twitter, Web Scraping & More

Book a demo and supercharge your sales team

Today:

• OpenAI’s GPT-5 Progress Slows

• Gemini Model Faces Improvement Challenges

• OpenAI's New AI Agent

• Elon Massive Colossus Supercomputer

• OpenAI Unveils U.S. AI Strategy

• Perplexity Launches Advertising on Platform

OpenAI LEAK: "AI progress will HALT" | GPT-5 is Slowing Down

OpenAI’s Orion model didn’t improve as much as expected, raising questions about simply making AI bigger. Experts worry there might be a limit to how smart AI can get by just increasing size and data. 

OpenAI is now focusing on enhancing models after their initial training using methods like reinforcement learning, where AI learns through trial and error, and human feedback to make it safer and more useful. They are also exploring specialized AI tools instead of one large model. 

WATCH THE VIDEO ON YOUTUBE

Gemini Model Faces Improvement Challenges

Google is facing challenges in improving its Gemini AI model as quickly as before, similar to OpenAI’s recent struggles. Despite using more data and computing power, Gemini hasn’t advanced as expected, questioning the belief that bigger models automatically become smarter. 

Google is now exploring new methods, such as adjusting how the AI learns and using feedback from users, to enhance its models. They are also developing specialized AI tools instead of one large model. Additionally, Google is investing heavily in better data and faster response times. This shift shows that improving AI may require more than just increasing size and data.

OpenAI's New AI Agent

OpenAI is set to release an autonomous AI agent named “Operator” in January, initially as a research preview and developer tool, enabling independent computer control and task performance. Competition in developing AI agents is heating up, with Anthropic and Google also working on their versions with advanced capabilities like “computer use,” positioning AI systems to actively interact with interfaces beyond text and images.

The consumer release date for Operator remains undisclosed, but its development signifies a shift towards AI systems capable of engaging with computer interfaces, hinting at a significant breakthrough in AI technology. Major AI companies are focusing on autonomous AI agents, aiming to commercialize these models to justify the substantial investment in AI development, with expectations that these agents will be the next ground-breaking product in the industry.

Elon Massive Colossus Supercomputer

Elon Musk surprised competitors by quickly building a massive supercomputer, named Colossus, for his AI company xAI. Located in a repurposed factory in Memphis, it uses 100,000 powerful graphics chips to train artificial intelligence, much larger and faster than rivals like OpenAI. 

Musk finished the project in just four months by cutting usual steps, such as securing enough power beforehand. This rapid development has worried other tech giants, leading them to rush their own data center expansions. Musk’s bold approach challenges industry standards, sparking a fierce competition to create the most powerful AI systems.

OpenAI Unveils U.S. AI Strategy

OpenAI is unveiling a detailed plan for the U.S. to strengthen its artificial intelligence (AI) infrastructure and form an alliance to rival China. The plan includes creating special AI zones, boosting government projects with private funds, and building new energy sources like solar, wind, and nuclear power. 

OpenAI aims to establish data centers in the Midwest and Southwest, areas with plenty of land for renewable energy and agriculture data. They also propose a national network to improve power and internet connections. This strategy seeks to create jobs, grow the economy, and ensure the U.S. leads in AI technology.

Perplexity Launches Advertising on Platform

Perplexity is starting to show ads in the U.S. as “sponsored follow-up questions” like “How can I use LinkedIn to enhance my job search?” These ads appear beside answers and are labeled “sponsored,” featuring brands such as Indeed and Whole Foods. Perplexity uses advertising to create a steady income since subscriptions alone aren't enough. They assure that ads won’t alter the AI-generated answers, maintaining accuracy and trust. 

Unlike OpenAI and Google, which have different ad approaches, Perplexity targets educated, high-income users. However, the company faces challenges, including plagiarism lawsuits and doubts about its ad effectiveness.

🧠RESEARCH

JanusFlow: Harmonizing Autoregression and Rectified Flow for Unified Multimodal Understanding and Generation

JanusFlow is a new system that combines understanding and creating images using one model. It merges language tools with advanced image techniques without complicated changes. By separating and aligning parts for understanding and generation, JanusFlow performs as well or better than specialized models, making image and language tasks more efficient.

BLIP3-KALE: Knowledge Augmented Large-Scale Dense Captions

BLIP3-KALE is a large collection of 218 million image and text pairs. It combines detailed computer-made descriptions with real-world captions to create accurate image explanations. Using advanced models, it improves how computers understand and describe images. This dataset helps train smarter systems for tasks involving both images and language.

Watermark Anything with Localized Messages

"Watermark Anything with Localized Messages" introduces WAM, a smart tool that hides secret messages in specific parts of images without noticeable changes. It can detect and retrieve these messages even if the image is edited or altered. WAM works well on high-quality pictures and can manage multiple hidden messages in small areas.

Rapid Response: Mitigating LLM Jailbreaks with a Few Examples

"Rapid Response" enhances the safety of large language models by quickly stopping harmful attempts after seeing only a few examples. The researchers created a test to evaluate these defenses and found their best method greatly reduced successful attacks. This approach effectively helps prevent the misuse of powerful language models.

Artificial Intelligence, Scientific Discovery, and Product Innovation*

A study at a U.S. company gave AI tools to 1,018 scientists. Those using AI discovered 44% more materials, filed 39% more patents, and increased product innovation by 17%. Top researchers benefited the most by effectively using AI, while others saw little improvement and felt less satisfied with their work.

🛠️TOP TOOLS

AI Search Grader - See how visible your brand is in AI-powered search engines.

Omnifact - Privacy-first generative AI platform made for the workplace

Twill - Accelerating the delivery of digital-led care

Trellis - AI engine transforms complex data sources - like financial documents, voice calls, and emails - into structured SQL-ready format for use by data and ops teams.

Kypso - Automate and scale your R&D teams with confidence, integrated into the apps you use for work.

📲SOCIAL MEDIA

🗞️MORE NEWS

• China’s first full-size electric humanoid robot, Tiangong, is now open-source. This allows businesses and researchers to use its design, reducing development time and costs. The move aims to integrate robots into daily life faster and more affordably.

• YouTube is experimenting with AI to let creators restyle licensed songs for shorts. Creators can modify elements like mood or genre, generating a unique 30-second soundtrack. AI-generated voices from artists like Charlie Puth and Charli XCX can be used in the remixes.

• DeepL, a German translation company, launched DeepL Voice for real-time text translations of spoken words in meetings and videos. It supports multiple languages, shows translated text instantly, and ensures user data privacy without storing voices.

• Exo Labs lets users run powerful open-source AI models on Apple’s new M4 Macs by linking multiple devices. This lowers costs, improves privacy, and gives individuals and businesses control by using AI locally instead of expensive cloud services.