Google has published LangExtract, a Python library that converts unstructured text into audit-ready tables while linking each extracted fact to its original source. The tool, released as open source, gives reviewers an auditable trail and includes interactive visualisations to verify large-language-model outputs. The move comes as developers unveil a slate of other free, community-maintained alternatives to commercial AI builders and knowledge-management services. Dyad, billed as a “full-stack” AI application builder, and Open DeepWiki, which turns any software repository into a multilingual, searchable wiki, have both been posted to public code repositories under the Apache 2.0 licence. Open DeepWiki’s maintainers say the software can run locally and plug into a range of language models, including OpenAI’s GPT-4, Google’s Gemini and the open-source Ollama family. Early adopters say the new projects replicate or surpass features found in proprietary platforms such as Lovable, V0 and Replit, but without usage fees or data leaving the user’s machine. The releases highlight a growing shift toward openly licensed developer tools aimed at lowering the cost and privacy barriers to building AI-powered applications.