The most important news pieces from the AI world are here for you, whether you’re an AI enthusiast or just wish to keep an eye on what’s going on.
Focus Topic: ChatGPT just became the new App Store – and small entrepreneurs should pay attention
OpenAI’s newest update to ChatGPT didn’t make major headlines, but it probably should have. Developers can now submit their own apps directly to the ChatGPT App Directory – an in-chat marketplace where users can discover and use AI-native tools in real time. While this might seem like a feature aimed at large platforms or enterprise developers, the real opportunity lies elsewhere: small teams and early-stage startups now have a direct path to distribution inside ChatGPT.
Until now, getting users to discover your AI tool often meant building landing pages, climbing Google’s SEO ladder, or hoping someone mentions you in a subreddit or newsletter. With the App Directory, any builder can now integrate their tool into ChatGPT and let users trigger it naturally in conversation, just by mentioning the app or selecting it from a menu. It’s not just visibility, it’s utility embedded at the point of intent.
For example, a startup offering resume feedback, real estate filtering, PDF search, or grocery ordering can now build a ChatGPT-native interface and let users execute the full workflow without leaving the chat. Think of it like a connector: instead of asking people to visit your app, you meet them where their attention already is. And with OpenAI experimenting with context-aware recommendations, your app could even be suggested mid-conversation if it’s relevant to the user’s need.
Right now, monetization is limited – apps can link out to complete transactions – but OpenAI has confirmed that in-app purchases and digital goods are on the roadmap. In other words, this may be the early phase of an entirely new app economy. And unlike the early days of mobile, where distribution was locked behind Apple and Google, ChatGPT apps are more open, deeply contextual, and AI-native from the start.
If you’re building something useful, there’s now a direct line to your users – through the world’s most popular AI interface.
LLMs & AI Models
- OpenAI released GPT-5.2-Codex, a new model for code generation with improved programming abilities and enhanced cybersecurity features.
- GPT Image 1.5 launched, generating images 4x faster and improving text, lighting, and facial accuracy – now top-ranked in image generation tests.
- OpenAI introduced the FrontierScience benchmark, a new test for scientific reasoning tasks, where GPT-5.2 scored highest.
- Mistral released OCR 3, a model for reading scanned documents and notes, achieving best-in-class results in OCR benchmarks.
- Google launched Gemini 3 Flash, a model that’s 3x faster and 4x cheaper than its predecessor, matching GPT-5.2-level results.
- Nvidia introduced Nemotron 3, a family of open models optimized for multi-agent AI systems, along with datasets and training tools.
- Zoom’s federated AI system beat Gemini in the Humanity’s Last Exam test, showing how multi-model integration can outperform single models.
- ChatGPT now supports personality settings, allowing users to adjust tone, formality, emoji use, and assistant style directly. Read more in my post here.
New Tools
- xAI launched Grok Voice Agent API, enabling real-time, multilingual voice agents for apps and cars, priced at $0.05/minute.
- Meta introduced SAM Audio, a model that isolates sounds from videos and audio using text, visual clicks or timeline inputs.
- Google Labs revealed CC assistant, a Gemini-based tool that connects to Gmail, calendar and files to offer personalized daily briefings. Currently on waitlist.
- Black Forest Labs launched FLUX.2 [max], its most powerful image model yet, offering real-time visuals from web-based data.
- Manus 1.6 adds Max agent and Design View, boosting autonomy and mobile support, while enabling visual creation and editing.
- Google added real-time speech translation to any Android earbuds, supporting 70+ languages and maintaining tone and rhythm.
👉 Explore these tools: Grok Voice Agent | SAM Audio | Google CC waitlist | FLUX.2 | Manus 1.6
Other Quick Picks
- AI evolves unevenly, excelling at hard tasks but failing at basic ones like memory and visual precision, says Ethan Mollick.
- US Energy Department launched Genesis Mission, partnering with 24 tech giants including OpenAI and Nvidia to accelerate nuclear and quantum research.
- Figure AI’s CEO launched Hark, a new AI lab focused on reasoning, backed by $100M in personal funding.
- Lovable raised $330M, valuing the AI coding platform at $6.6B.
- Elon Musk said xAI must survive 2-3 years to win the AI race, citing data center growth and $20-30B in annual funding potential. And also expecting AGI in 2026.
- Amazon considers investing $10B in OpenAI, signaling a shift away from Microsoft exclusivity and a deeper AWS partnership.
- Stanford predicts 2026 will be a reality check, as firms start measuring real AI impact and move beyond AGI hype.
- HubSpot’s CEO says SEO is dying, and content must now be optimized for LLMs via Answer Engine Optimization, not search engines.
- Multi-agent AI setups don’t always help, improving finance tasks but failing in step-by-step logic problems, says new Google-MIT study.
- AI models now pass all CFA levels, with Gemini 3.0 Pro scoring 97.6% on Level 1, showing AI’s growing impact in finance.
- AI agents are mostly used for deep knowledge work, not simple automation, says Perplexity-Harvard user study.
🇪🇪 AI News from Estonia
- Only 14% of Estonian companies used AI in 2024, far behind Denmark, due to lack of skills, poor data quality and unclear use cases.
- EasyRef wants to make referrals the new standard, offering a platform where local businesses gain customers via reward-based recommendations.
- Tarmo Koppel compares AI to fire and the wheel, saying it will boost productivity and change the economy, while humans focus on strategy and creativity.
- Laura Niit warns against blindly trusting AI in accounting, citing risks of legal errors and fines due to lack of control and expertise.
🎙️ AIPowerment Podcast dropped a new episode featuring a 2025 AI year-end review, where me and Sandra discuss major trends, tools, surprises, and lessons from the past year.
🎧 Listen to AIPowerment Podcast on Spotify, Apple Podcasts, and YouTube.
Want to stay in the loop?
Straight to your inbox: practical AI updates, finance use cases, tools to try, and upcoming trainings.
I cut the noise and send only what’s actually useful.
💼 Exploring how to make AI work in finance? Let’s talk practical use cases – connect with Gerlyn Tiigemäe for expert guidance.
📚 If you would like to participate in one of my trainings or listen to speaking engagements, here are the upcoming ones:
- Training “AI võimalused finantsvaldkonnas” @ Äripäeva Akadeemia, last free spots for December – information
- Training “AI võimalused projektijuhtimises” @ Äripäeva Akadeemia, next trainings in 2026 – information
- Training “AI assistendi roll ja võimalused”, 03.02.2026 – information
