Kosmos
Kosmos is the next-generation “AI Scientist” developed to perform autonomous discovery by reading vast amounts of scientific literature and executing code to reach novel conclusions. It uses structured world models to efficiently incorporate information gathered over hundreds of agent trajectories and maintain coherence throughout tens of millions of tokens, thereby transcending the context-length limits of earlier language-model-based tools. A typical Kosmos run might read about 1,500 papers and execute 42,000 lines of analysis code, enabling it to perform in one day what beta users estimated would take a human scientist six months. Its outputs are fully traceable; each conclusion in a Kosmos report can be linked to the specific lines of code and passages in the literature that inspired it, allowing for full auditability of its reasoning.
Learn more
FutureHouse
FutureHouse is a nonprofit AI research lab focused on automating scientific discovery in biology and other complex sciences. FutureHouse features superintelligent AI agents designed to assist scientists in accelerating research processes. It is optimized for retrieving and summarizing information from scientific literature, achieving state-of-the-art performance on benchmarks like RAG-QA Arena's science benchmark. It employs an agentic approach, allowing for iterative query expansion, LLM re-ranking, contextual summarization, and document citation traversal to enhance retrieval accuracy. FutureHouse also offers a framework for training language agents on challenging scientific tasks, enabling agents to perform tasks such as protein engineering, literature summarization, and molecular cloning. Their LAB-Bench benchmark evaluates language models on biology research tasks, including information extraction, database retrieval, etc.
Learn more
Perplexity Computer
Perplexity Computer is an AI-powered super agent designed to autonomously complete complex digital tasks from start to finish. Users simply describe the outcome they want, and the system breaks the request into structured subtasks executed by specialized AI models. It can build websites, generate reports, compile datasets, and create multimedia content with minimal manual input. The platform dynamically selects the most suitable AI models for each component of a project, optimizing for research, images, video, or quick searches. Designed for extended autonomous operation, it can run workflows for hours or longer without interruption. By abstracting away technical complexity, it transforms high-level intent into fully executed results. Perplexity Computer streamlines advanced AI capabilities into a single, outcome-focused interface.
Learn more
Acade
Acade is an AI research co-scientist who starts with a research question and turns it into a structured, verifiable research loop. It helps researchers map literature, propose traceable hypotheses, plan experiments, interpret results, and turn the full path into an evidence-backed report while keeping the scientist in control. It is built for human-in-the-loop research, supporting users as they search, compare, critique, and document evidence without replacing scientific judgment. Acade begins with research question intake, capturing the domain, goal, constraints, files, assumptions, and expected decision before the agent starts. It can organize relevant papers, claims, methods, debates, and research gaps into a literature-grounded map while preserving source provenance. It also generates hypothesis cards that compare evidence, counter-evidence, novelty, feasibility, and risk, helping researchers review candidate ideas before execution.
Learn more
