# forck.live — 13–19 January 2025 > AI lab announcements, news reports and podcasts. forck.live follows what companies announce about their own products. Press may corroborate an official story; it never leads and never appears as an item of its own. A missing date stays missing. A title is the publisher's words — where a company does not publish in English, an English rendering is listed separately as `Title in English` and named as this site's translation, never as theirs. This file is one issue of the weekly archive: every confirmed first-party announcement of ISO week 2025-W03, 13–19 January 2025, newest first. The window never moves, so this URL always names the same seven days — but a post published inside them and detected later is added when it arrives, so a closed week can still gain entries. Every entry is one first-party announcement, and carries the same eight fields in this order before any others: Company · Announced · Category · Coverage · Announcement · Group · Source · Record `Source` is the company's own page the announcement was traced to and `Record` is this site's page for it. Both are safe to cite. `Coverage` is a number of outlets, or `not counted` — which means this record reached no count of its own, never that nobody wrote about it. `Category` is this tracker's own analysis, and is authoritative wherever it is stated. `not stated` means either that nothing has analysed the post yet or that the analysis placed it outside this taxonomy; neither is a verdict about the announcement. `Announcement` is whether the post is news of something the company did, rather than a how-to, a customer story, a changelog listing or a staffing note. `Group` is which of this record's groups the entry falls in, carried on the entry itself so it holds where the groups are headings and where they are not. One of: `models`, `covered`, `announcements`, `routine`. Anything after `Record` is held detail that only some announcements carry. - This week's page: https://forck.live/week/2025-W03 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Lovable — MySafe x Lovable: Hackathon Canada Winner - Company: Lovable (lovable.dev) - Announced: 2025-01-17T18:03:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://lovable.dev/blog/2025-01-17-mysafe-x-lovable-hackathon-canada-winner - Record: https://forck.live/items/12193-mysafe-x-lovable-hackathon-canada-winner - Subject: Lovable / AI app builder Lovable published a case study about a team of computer science students who built a community safety app called MySafe during a 20-hour Canadian hackathon using Lovable's AI-powered platform. The app features a real-time incident map and an AI chatbot for safety forecasts. The team overcame challenges with Lovable credits and Gemini integration to complete a working prototype. ### Lovable — How Prompt Father Built a Thriving Community with Lovable - Company: Lovable (lovable.dev) - Announced: 2025-01-17T17:05:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://lovable.dev/blog/2025-01-20-how-prompt-father-built-a-thriving-community-with-lovable - Record: https://forck.live/items/12194-how-prompt-father-built-a-thriving-community-with-lovable - Subject: Lovable / AI app builder An interview with Sancho (Prompt Father) describes how he uses Lovable to build MVPs and grow a community of developers and non-tech founders, including a $1,000 challenge for building with the platform. ### Lovable — From Idea to Impact: Yannis' Journey with Lovable - Company: Lovable (lovable.dev) - Announced: 2025-01-17T15:13:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://lovable.dev/blog/2025-01-17-from-idea-to-impact-yannis-journey-with-lovable - Record: https://forck.live/items/12195-from-idea-to-impact-yannis-journey-with-lovable - Subject: Lovable / AI app builder Lovable's customer story profiles Yannis, a digital marketer who built the micro-SaaS PrintPigeon in about three days using Lovable, spending $38 in credits. The piece covers his go-to-market approach, which combined Google Ads with programmatic SEO landing pages, and notes that nearly 50% of users were expats and that a seasonal Santa letter service drove unexpected traffic. It also recounts his path to joining the Lovable team and his advice on prompts, integrations, and solving real problems. ### Replicate — Generate short videos with the Replicate playground - Company: Replicate (replicate.com) - Announced: 2025-01-17 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replicate.com/blog/generate-videos-with-playground - Record: https://forck.live/items/2504-generate-short-videos-with-the-replicate-playground - Subject: Platform Replicate published a guide showing how to use its existing playground interface to generate short AI videos, starting from an image and using models like minimax/video-01-live and zsxkib/mmaudio for sound. The playground, originally built for image models, now supports video models as well. ### Replit — AI Coding is Better with Python - Company: Replit (replit.com) - Announced: 2025-01-16T15:04:02+00:00 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://replit.com/blog/ai-coding-is-better-with-python - Record: https://forck.live/items/17979-ai-coding-is-better-with-python - Subject: Replit Agent "Can everyone really code with AI?" I hear this question all the time. ### Lovable — The Lovable Prompting Bible - Company: Lovable (lovable.dev) - Announced: 2025-01-16T11:02:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://lovable.dev/blog/2025-01-16-lovable-prompting-handbook - Record: https://forck.live/items/12196-the-lovable-prompting-bible - Subject: Lovable / AI app builder Lovable published a handbook on prompt engineering, covering techniques such as meta prompting, reverse meta prompting, and structured prompt formats to improve AI output quality and efficiency. The guide includes a prompt library and advice on starting new projects, using diff and select, locking files, and design. ### LG AI Research — LG AI Research's Take on CES 2025 - The Dawn of an Industry-Specific AI Era - Company: LG AI Research (lgresearch.ai) - Announced: 2025-01-16 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=516 - Record: https://forck.live/items/11015-lg-ai-research-s-take-on-ces-2025-the-dawn-of-an-industry-specific-ai-era - Subject: EXAONE LG AI Research attended CES 2025 and concluded that the show marked the beginning of an era of industry-specific AI, with companies moving beyond generative AI to showcase AI agents and domain-specific solutions in areas such as manufacturing, finance, transportation, and beauty. LG Electronics demonstrated AI-applied products including the Gram laptop with an on-device AI feature called gramChat On-Device, powered by LG AI Research's EXAONE model, which can analyze and summarize documents, support English-Korean translation, and operate without a network connection. ### Mistral AI — Purr-fectly informed - Company: Mistral AI (mistral.ai) - Announced: 2025-01-16 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://mistral.ai/news/mistral-afp - Record: https://forck.live/items/3334-purr-fectly-informed - Subject: Mistral / Le Chat Mistral AI announced a product titled 'Purr-fectly informed' on January 16, 2025. ### Hugging Face — Introducing multi-backends (TRT-LLM, vLLM) support for Text Generation Inference - Company: Hugging Face (huggingface.co) - Announced: 2025-01-16 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/tgi-multi-backend - Record: https://forck.live/items/1769-introducing-multi-backends-trt-llm-vllm-support-for-text-generation-inference - Subject: Platform Hugging Face announces multi-backend support for Text Generation Inference (TGI), allowing integration with vLLM, TensorRT-LLM, llama.cpp, AWS Neuron, and Google TPU backends through a unified frontend layer. ### Hugging Face — Timm ❤️ Transformers: Use any timm model with transformers - Company: Hugging Face (huggingface.co) - Announced: 2025-01-16 - Category: developer-tool-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/timm-transformers - Record: https://forck.live/items/1768-timm-transformers-use-any-timm-model-with-transformers - Subject: Platform Hugging Face announces TimmWrapper, a tool that integrates any timm model with the Transformers library, enabling pipeline API support, quantization, fine-tuning, and torch.compile. ### OpenAI — Partnering with Axios expands OpenAI’s work with the news industry - Company: OpenAI (openai.com) - Announced: 2025-01-15T03:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/partnering-with-axios-expands-openai-work-with-the-news-industry - Record: https://forck.live/items/653-partnering-with-axios-expands-openai-s-work-with-the-news-industry - Subject: GPT / ChatGPT / API OpenAI announces that publishers representing hundreds of newsrooms and content brands are using OpenAI partnerships and grant programs to adopt AI tools and strengthen the news ecosystem, while ChatGPT users gain access to information from leading, reliable publications. ### Hugging Face — Train 400x faster Static Embedding Models with Sentence Transformers - Company: Hugging Face (huggingface.co) - Announced: 2025-01-15 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/static-embeddings - Record: https://forck.live/items/1770-train-400x-faster-static-embedding-models-with-sentence-transformers - Subject: Platform Hugging Face introduces a training method for static embedding models that are 100x to 400x faster on CPU than state-of-the-art models while retaining about 85% of performance. They release two models (for English retrieval and multilingual similarity), training scripts, and datasets. ### Sakana AI — Transformer²: Self-Adaptive LLMs - Company: Sakana AI (sakana.ai) - Announced: 2025-01-14T15:00:00+00:00 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://sakana.ai/transformer-squared - Record: https://forck.live/items/4724-transformer-self-adaptive-llms - Subject: Sakana models Sakana AI announces a research paper titled Transformer², which proposes a self-adaptive machine learning system that dynamically adjusts its weights for various tasks using Singular Value Finetuning (SVF) and reinforcement learning, outperforming static approaches like LoRA. ### OpenAI — Adebayo Ogunlesi joins OpenAI’s Board of Directors - Company: OpenAI (openai.com) - Announced: 2025-01-14T09:00:00+00:00 - Category: funding-company-news - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/adebayo-ogunlesi-joins-openais-board-of-directors - Record: https://forck.live/items/654-adebayo-ogunlesi-joins-openai-s-board-of-directors - Subject: GPT / ChatGPT / API Adebayo Ogunlesi has joined OpenAI's Board of Directors. ### LG AI Research — [NeurIPS 2024 Series] Introduction to Mutual Information Estimation Benchmark Research - Company: LG AI Research (lgresearch.ai) - Announced: 2025-01-14 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/blog/view?seq=514 - Record: https://forck.live/items/4813-neurips-2024-series-introduction-to-mutual-information-estimation-benchmark - Subject: EXAONE LG AI Research presented a paper at NeurIPS 2024 in the Datasets & Benchmark track introducing a benchmark suite for evaluating neural mutual information estimators on unstructured datasets like images and text, addressing limitations of existing Gaussian-based benchmarks. ### Kyutai — Announcing Helium-1 Preview - Company: Kyutai (kyutai.org) - Announced: 2025-01-13 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://kyutai.org/blog/2025-01-13-helium - Record: https://forck.live/items/18359-announcing-helium-1-preview - Subject: Moshi / Pocket TTS Kyutai released a preview of Helium-1, a lightweight multilingual language model designed for edge and mobile devices. ### Mistral AI — Codestral 25.01 - Company: Mistral AI (mistral.ai) - Announced: 2025-01-13 - Category: new-model - Coverage: not counted - Announcement: yes - Group: models - Source: https://mistral.ai/news/codestral-2501 - Record: https://forck.live/items/3335-codestral-25-01 - Subject: Mistral / Le Chat Mistral AI announces Codestral 25.01, a research model. ### Hugging Face — AI Agents Are Here. What Now? - Company: Hugging Face (huggingface.co) - Announced: 2025-01-13 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/ethics-soc-7 - Record: https://forck.live/items/1771-ai-agents-are-here-what-now - Subject: Platform Hugging Face blog post overviewing AI agents, discussing ethical tradeoffs, and recommending against fully autonomous AI agents.