"Kosmos: An AI Scientist for Autonomous Discovery"
Kosmos is an AI scientist tool from Edison Scientific. It runs for up to 12 hours & uses 200 agent rollouts, collectively executing an average of 42,000 lines of code & reading 1,500 papers per run. Users reported that a single Kosmos run performed the equivalent of 6 months of their own research time on average. Kosmos can propose novel analytical approaches.
ekoAcademia is a search engine for Arxiv papers. It automatically creates 2.5 minute audio podcast summaries of each paper that you select in the search results if nobody has browsed that article yet. The created audio summaries are saved for future users. The alphaXiv tool ( https://www.alphaxiv.org ) is an alternative Arxiv search engine that creates AI audio summaries that are twice as long.
AI Paper Finder search engine
One advantage of this tool is that you can search for some papers that are listed even though they are still under review for conferences taking place in 2026 (eg. ICLR 2026). You can specify the number of search results you want and export them all into a text format file.
"Tongyi DeepResearch Technical Report"
Tongyi DeepResearch, an agentic LLM, is specifically designed for long-horizon, deep information-seeking research tasks.
Tongyi is open-source and has some advances that make it more like a 2nd generation deep research tool.
https://arxiv.org/html/2510.24701v1
Tongyi got me interested in 2nd generation deep research tools so I had AI2's Asta 2nd generation deep research tool generate a report on them:
https://asta.allen.ai/share/913a85a2-6309-40a5-b1b4-91312acc37df
Want a banner ad under your bash prompt? Probably not, but Sourcegraph just made the world’s first ad-supported AI coding agent.
With Amp Free, you toggle a setting and — voilà — ads appear at the bottom of your editor, while Amp handles your code reasoning and edits. In exchange, you skip the subscription fee and opt into “Training Mode,” which shares your thread data for model improvement.
Read all about Amp Free here: https://ainativedev.co/znc
"DeepTRACE: Auditing Deep Research AI Systems for Tracking Reliability Across Citations and Evidence"
Article compares the performance of GPT-4.5/5, You.com, Perplexity, Copilot/Bing, and Gemini. Our findings show that generative search engines and deep research agents frequently produce one-sided, highly confident responses on debate queries and include large fractions of statements unsupported by their own listed sources.
instawarp is a new search engine that allows you to search multiple search engines at one time. You need to allow instawarp to open popup windows in your browser, so that the search results for each search engine will each open in different tabs in your browser. If you set up a free login account, you can add more search engines to the 10 default ones. Some of the search engines are AI tools like ChatGPT or Perplexity.
#MCP is live in #openQA. #openSUSE now supports #AI-friendly, read-only access to test data & paves the way for #open, intelligent #automation. #Testing #AItools https://news.opensuse.org/2025/09/23/openqa-gains-ai-friendly-interface/
The newest release of Firefox web browser, v144.0, has new AI features. When you right click on a webpage (see image below), a new option that you are offered is "Ask an AI Chatbot." Selecting that option gives you a choice of 4 different AI chatbots to use. I did a little informal testing and Google Gemini was by far the best chatbot for questions like asking it to summarize the webpage you are looking at. I noticed this new feature on Firefox running on Windows 11.
The #FTC Is Disappearing Blog Posts About #AI Published During Lina Khan’s Tenure
The Federal Trade Commission removed several blog posts in recent months about open source and potential risks to consumers from the rapid spread of commercial #AItools.
#opensource #linakhan
https://www.wired.com/story/ftc-removes-blog-posts-about-ai-authored-by-by-lina-khan/
Dean Giustini posted a great article on X about a Knowledge Synthesis (KS) & Artificial Intelligence (AI) Search Wiki 2025 that he created. The wiki has some really great content related to AI tools in those areas.
BioMedSearch, a multi-source AI biomedical information retrieval framework based on LLMs. The method integrates literature retrieval, protein database and web search access to support accurate and efficient handling of complex biomedical queries. For needle in the haystack type queries, AI2's Asta tool https://asta.allen.ai is a much better tool.
Search tool:
https://bio-med-search.com/
Paper on this tool:
https://ui.adsabs.harvard.edu/abs/2025arXiv251013926L/abstract
#research #medicine #AItools #science
Ribbit Ribbit is a pretty useful tool summarizing top arxiv papers, selected by a machine learning pipeline. It also offers an option to listen to a AI generated podcast summary of the paper. It is one of the best current awareness tools of this type that I have seen,
Documentation:
https://ribbitribbit.co/md/home.md
"Comparison of Elicit AI and Traditional Literature Searching in Evidence Syntheses Using Four Case Studies"
Cochrane Evid Synth Methods, 9-27-25
The sensitivity of Elicit was poor, averaging 39.5% compared to 94.5% in the original reviews. Elicit identified some included studies not identified by the original searches & had an average of 41.8% precision which was higher than the 7.55% average of the original reviews.
Many #workers are using unapproved #AItools at #work - and sharing a lot of #privatedata they really shouldn't
getvizz - AI-powered repository analysis that turns complex codebases into interactive documentation, dependency graphs, and intelligent conversations.
The example shown in the left image is looking at the graph for https://github.com/NirDiamant/RAG_Techniques .
This is a great article by Nicolas Bustamante on some of the technical limitations of RAG as a search algorithm.
"The RAG Obituary: Killed by Agents, Buried by Context Windows"
Why Retrieval-Augmented Generation Won’t Survive the Context Revolution and the End of Chunking, Embeddings, and Rerankers as We Know Them.
https://www.nicolasbustamante.com/p/the-rag-obituary-killed-by-agents
Futurescape Libraries AI Toolkit
This 148 page toolkit is designed for leaders of research libraries who want to integrate the ARL/CNI 2035 AI scenarios
into ongoing planning activities—such as annual operating plans, strategic reviews, or cross-institutional conversations.
Wikidata Vector Database
The Wikidata Vector Database, part of the Wikidata Embedding Project , stores semantic embeddings of Wikidata entities to enable context-aware search using vector similarity. It uses JinaAI and at this point the response time for queries is rather slow.
Need translations handled in bulk across thousands of spreadsheet rows? (Un)Perplexed Spready with Perplexity AI supports contextual translation for 30+ languages.
https://matasoft.hr/qtrendcontrol/index.php/un-perplexed-spready/un-perplexed-spready-various-articles/148-matasoft-s-ai-driven-spreadsheet-processing-services-transforming-data-workflows
#TranslationAI #Localization #MultilingualData #UniversalAI #SmartData #AItools #Automation #ResearchTools #TextMining #BItools #DataPipeline #SmartIntegration #ExcelTips #SmartTools #AIComparison #AIDriven #TextAnalysis #SpreadsheetAI #DataProcessing #AI #DataProcessing
