Hi everyone, I'm a master's student in Germany and spend most of my week on research. Most of that time went into moving things between tools: a reference manager, a spreadsheet, a survey tool, a transcription service, and finally LaTeX. Every handoff lost context, and once AI entered the workflow it got worse. A number or a citation would show up in a draft and I could no longer say where it came from. SixSentences puts those steps in one workspace: literature search and screening, research data and figures, surveys and text or spoken interviews with explicit consent and retention gates, and a LaTeX writing environment. What matters is what survives between them. Sources, transformations and decisions stay inspectable, so a figure or a citation in the manuscript can be traced back to the record it came from. AI steps are optional and run against the providers, budgets and data-processing rules the operator sets. The screening models are fine-tuned transformer encoders I trained myself, not a prompt. It is Apache-2.0 and self-hostable: clone, make up, done. This is an early alpha, so APIs and UI can still change, and I would much rather hear about the rough edges now than after 1.0. Which part of your research workflow is currently the least traceable?

SixSentences_

Open AI research workspace where every claim is traceable

Open AI research workspace where every claim is traceable