Corpus is an AI document Q&A product for teams: upload PDFs and web pages into shared workspaces, ask questions in natural language, and get answers pinned to the source passages they came from. Its distinguishing views compare answers across documents side by side and automate repeat analysis as workflows.[1]
Quick verdict
Teams doing cited research across PDF and web collections get the most from Corpus: shared workspaces, source-linked answers, cross-document comparison, and exportable workflows cover the core literature-review loop. The decisive constraint is newness: 13 stars and a mid-2026 last push mean adopters bet on an early AGPL project with moderate self-hosting effort and unobserved hosted pricing, so critical work should keep an established fallback until Corpus matures.[1][2]
Key features
Natural-language questions over uploads Natural-language questions over uploads: ask plain questions against uploaded PDFs and web pages instead of writing search queries.[1]
Answers with source citations Answers with source citations: each answer links back to the source text passages it came from, so claims stay checkable.[1]
Workspaces and document sets Workspaces and document sets: organize documents into shared workspaces and sets so a team interrogates the same collection.[1]
Cross-document comparison Cross-document comparison: the cross-correlation view compares answers across documents, which suits due-diligence and research contrast work.[1]
Automated workflows Automated workflows: repeat analysis can be created as workflows rather than re-asked by hand each time.[1]
Workflow output and export Workflow output and export: workflow results can be viewed and exported elsewhere instead of staying locked in the app.[1]
Multi-provider models Multi-provider models: works with models from OpenAI, Anthropic, Google, and xAI, with OpenAI and Voyage AI embeddings.[1]
Docker Compose self-hosting Docker Compose self-hosting: one compose file starts the frontend, API, database, search, queue, orchestration, cache, and storage services.[1]
Who will love it
Teams researching across document collections: workspaces plus cited answers fit literature-style review over PDFs and saved pages.[1]
Reviewers comparing sources: the cross-document comparison view serves due-diligence contrast where answers must stay attributable.[1]
Self-hosting teams comfortable with multi-service Docker: the AGPL codebase and compose path suit operators who already run Postgres, search, and queue infrastructure.[1][2]
Who should avoid it
Anyone needing a proven product: 13 stars, a January 2026 creation, and a June 2026 last push mark a very early project, so adopters carry new-project risk.[1]
Teams wanting light operations: eight coordinated services plus provider model configuration make this moderate operator work, not casual self-hosting.[1]
Buyers needing clear hosted pricing: no public plan or price is observable on the official site, so budget approval has nothing firm to cite.[1][2]
Alternatives and compatibility
The AGPL-3.0 license keeps a fork path open: a team can preserve and adapt the primitives independently of the hosted product's future.[1]
Workflow output exports elsewhere, so analysis is portable and leaving does not strand finished work inside the app.[1]
No catalogue alternative edge is evidenced yet, so treat Corpus as an early-stage bet: pair critical work with an established cited-answer tool until the project matures.[1][2]
Questions worth asking
What is the best Corpus alternative for non-technical users?
Is Corpus really no-code?
No. The primary path is Docker Compose across eight services plus provider model configuration, and running without Docker needs Python, Node, Poetry, Alembic, and each infrastructure service started individually.[1]
How much does Corpus cost per month?
Sources and editorial review
Source research. Reviewed on October 9, 2026. Written by Sanjib Sah. AI assisted; reviewed by a human editor.
We have not run or tested this product.
- Corpus README (official)
official · Observed October 9, 2026
Repository README: cited document Q&A, workspaces and document sets, cross-correlation and workflow demos, Docker Compose and non-Docker setup, full service architecture, supported model and embedding providers, AGPL-3.0.
- noetic-sys/corpus repository (official)
official · Observed October 9, 2026
GitHub repository noetic-sys/corpus: Open Document Intelligence, organization owner, Python/TypeScript stack, star and recency identity facts.
- noetic-sys/corpus API metadata (official)
official · Observed October 9, 2026
Repository metadata: created 2026-01-02, last push 2026-06-26, 13 stars, not archived, AGPL-3.0 license record.
- Corpus official site (official)
official · Observed October 3, 2026
Official site evidences a hosted product at onecorpus.com but rendered as an unreadable JavaScript shell with no pricing or feature detail.
Compared with these products
No explicit proprietary baseline is recorded.