Dialog is an open-source app for deploying retrieval-augmented generation (RAG) chatbots without deep API development work, hosted on GitHub with about 430 stars under the MIT license. Built by the Talkd team, it aims to let you spend less time coding and more time improving your model.
The project provides an API for deploying the LLM of your choice, structured around the companion dialog-lib library, and uses modern web and LLM frameworks including LangChain. A distinctive focus is “humanizing” RAG answers: keeping the answer scope tightly delimited to your material and making responses sound natural rather than robotic. Deployment is Docker-based, with Docker Compose for the first run, and documentation lives on the project’s site. Dialog is positioned for programmers who are interested in AI and want to put a RAG in front of users quickly, then iterate on retrieval quality and tone.
The team started with a specific frustration: many RAG chatbots answer too broadly and sound mechanical. Dialog therefore focuses on delimiting the answer scope to the material you provide and tuning responses to sound human, with an architecture that separates the API layer from the underlying dialog library so the retrieval and language behavior can evolve independently. A community Discord supports users deploying it.
Dialog is free and open source under the MIT license. You run it on your own infrastructure and only pay for whatever LLM provider you connect to it.
It is aimed at developers and small teams who want to ship a document-grounded chatbot quickly, especially those who care about controlling answer scope and tone, and who would rather configure a ready RAG stack than build one from scratch.