Open-source infrastructure for agents that keep improving from use
Emerged finds new products worth a look, daily — what’s new today
It automates the dull middle of ML research, and its three-model review step guards against an agent grading its own work.
AutoResearch from EvoMap is an open-source system that carries an AI/ML research idea through idea discovery, experiment planning, implementation, execution, analysis and review. It runs cheap pilots before scaling, keeps stateful experiments recoverable, and asks at least three distinct models to critique results independently. Every claim keeps provenance back to sources and critic reports. Apache-2.0, Python 3.10+, set up with a bring-up script and your own provider keys; it accompanies an arXiv paper (2608.17906).
Open-source infrastructure for agents that keep improving from use
Autonomous research system that evolves solutions across agent generations
A research workspace where the agent builds its own panes
Where engineers publish the AI tools and workflows they actually use
Agents that operate a desktop, a browser and a phone in one run
Open-source persistent AI teammates that use your tools like you do