Illustration · Files, names, messages and numbers on this board are fictional
SixSentences untangles this.
Allof this,in six sentences.
The scattered sources, disconnected data and drafts pinned around this card belong in one place. SixSentences connects them in one research workspace.
Registration paused
New registrations are currently paused. Existing users can sign in and continue their research.
Already have an account? Sign in
Prefer your own infrastructure? Self-host the community edition
Code availability
SixSentences_ is open source under Apache 2.0.8 Run the complete community stack on infrastructure you control, inspect how evidence moves through the workspace and help shape what comes next.
Quick start: git clone --depth 1 https://github.com/SixSentences/sixsentences.git, then cd sixsentences, then make up. Open http://localhost and sign in.
What make up does
- Writes a private local configuration
- Runs the fail-closed preflight
- Builds the images
- Starts the stack, health-gated
- Prompts once for the first owner account
Complete community stack
- Next.js workspace
- FastAPI service
- Research engine
- PostgreSQL + worker
- Browser capture
- macOS companion source
The community project is early alpha; APIs and UI contracts may still change.
Build it with usAppendix AQuestions
Where the evidence comes from, what the AI can get wrong and what happens to your data. The practical answers, without pretending research is automatic.
A.1What is SixSentences actually for?
SixSentences_ connects the work between a research question and a finished output: literature search, screening, source reading, data, interviews, figures and manuscript writing. The useful difference is the record around the answer: sources, decisions and changes stay connected instead of disappearing into a chat history.
A.2Is SixSentences open source?
Yes. The complete SixSentences_ community application stack is available on GitHub under the Apache 2.0 license, including the web workspace, API, research engine, worker, browser capture and macOS companion source. You can inspect it, contribute or operate it on your own infrastructure with the optional providers you choose. The hosted service, billing, production operations, marketing website and official signed native binaries are maintained separately. The community project is still early alpha, so APIs and UI contracts may change.
A.3Where do the papers and sources come from?
SixSentences_ can search OpenAlex scholarly metadata and, when enabled, the live web; follow citation relationships; start from known papers; and use records, PDFs and webpages you add. No index covers every field, so a rigorous review may still need discipline-specific databases, institutional access, hand searching or a librarian-designed strategy.
A.4Can I trust the citations and AI-written text?
Not without checking them. SixSentences_ is designed to connect claims to real source records and, where full text is available, to supporting passages. Models can still miss studies, misread evidence or draft unsupported text, so every material claim, number, quotation and citation should be verified against the original source.
A.5Will it run a systematic review for me?
It can preserve search routes, eligibility criteria, screening reasons, duplicate handling and flow counts for PRISMA-oriented reporting. It cannot make a search exhaustive or a review automatically compliant with PRISMA, a journal policy or a disciplinary standard. Method design and final study selection remain researcher decisions.
A.6Can it fetch paywalled papers for me?
It does not bypass paywalls or publisher access controls. Full text is attached only when it is openly available, supplied by you or otherwise lawfully accessible. If a source can be identified but no verifiable PDF is available, you can keep the citation and add a lawful copy yourself.
A.7What happens to my PDFs, prompts and research data?
Workspace content is not used to train SixSentences_ foundation models. Active workspace content is stored in Germany and encrypted recovery backups are stored in Amsterdam. A triggered AI feature may send only the material needed for that task to the processors described in the Privacy Policy and DPA; you remain responsible for the legal basis when uploading participant or third-party data.
A.8Can I export my work or delete it again?
Yes. Literature records can be exported as RIS, BibTeX or CSL JSON, and review, data and manuscript workspaces provide formats suited to their material, including source files and compiled documents. Saved content and the account can be exported or deleted through the available product controls; encrypted recovery copies expire under the documented backup rotation.
A.9How do the plans and research capacity work?
Free provides a one-time starter capacity for trying the core workflow. Scholar contains the complete individual workflow; Cohort adds a shared workspace, pooled capacity and parallel research runs. Capacity reflects the work behind searches, model calls, screening, extraction and generated outputs, so there is no fixed cost per click. AI-led interview studies are included with Scholar and Cohort. A spoken session temporarily reserves capacity for its authorized duration. After the server confirms the session has ended, recorded model costs determine the final usage relative to your plan budget, including the automatic analysis; unused reservations return. Duration is shown separately. If the server cannot confirm closure, the authorized cap may provisionally count. Contact support about an incorrect charge: we review available session records and correct unjustified charges, including errors caused by our systems. Paid monthly capacity renews on the billing date; purchased boosts do not expire and require an active paid plan to use.
A.10Can I sign up without an invitation?
When public registration is open, yes: you can create a Free account without an invitation or paid subscription, then verify your email address to activate it. During early access, request an invitation through the access form. If registration is paused, new accounts cannot be created; existing users can still sign in. The access section shows current availability. Paid plans are available only to customers with a residence or business and billing address in Germany; monthly plans can be cancelled for the end of the current billing period.