A closer look
at how NyayaDesk works.
A practical guide to the product, its boundaries and the decisions that remain with people.
Explore the capabilitiesFollow a matter through the workspace.
Understand the case before you draft.Bring petitions and annexures into one matter. Extract allegations, reconstruct chronology and trace each finding to its source.Draft a response the record supports.Develop the legal position, prepare a source-linked draft and inspect the checker’s findings before human approval.Research the law. Reuse relevant work.Read the authority behind a proposition and find related work in your organisation’s authorised matters.Know what is due, who owns it and why.Connect hearings, court directions and evidence requests to a responsible person, a date basis and the record of completion.How quality
is evaluated.
An evaluation method has been preregistered. Results are not yet published, so no measured accuracy claim is made here.
Retrieval, citation support, unsupported claims and checker behaviour need their own measurements. Answer correctness requires human annotation.
No published results at this reviewQuestions worth asking.
Is NyayaDesk a legal chatbot?
No. It is a matter-centred litigation workspace. Legal AI is one way to work with the record; evidence, strategy, drafts, hearings, compliance and audit remain structured parts of the product.
Does AI approve or file documents?
No. AI prepares and checks. People confirm facts, perform legal approvals, execute affidavits and file through the court registry. The software records those acts.
Does an independent check guarantee a correct answer?
No. A separate model family checks the work, but model agreement is not legal truth. Source review, unresolved findings and human accountability still matter.
Can a new user create a workspace immediately?
No. Workspaces are provisioned through authorised onboarding. Membership must already exist for sign-in; single sign-on does not automatically create it.
Does all AI processing stay in India?
No. The current inference design uses OpenRouter with zero-retention routing and may process data outside India. Embeddings use a separate self-hosted route. Discuss the actual deployment and subprocessors before onboarding.
Which languages are supported?
The current interface, AI answers, fact-confirmation questions and drafts are English-only. OCR can read English and Telugu documents. A Telugu interface is deferred.
Are court integrations automatically available?
No. Availability depends on the source, adapter configuration and access. Manual entry remains important. Do not assume access to CAPTCHA-protected sources or a complete live court feed.
Are Word, contracts and e-signing included?
The Word round trip, NyayaDesk for Word, the contract register and playbook review are part of the product; confirm what your deployment enables during onboarding. Signing through NyayaDesk is not available; signed-copy verification checks a signed PDF and is not an e-signing service.
What is still ahead.
A Telugu interface, SAML, SCIM and signing through NyayaDesk are not available. Contract negotiation, clickwrap and e-stamping are outside the product's scope.
Source coverage and court integrations depend on the enabled adapters and access. Confirm specific requirements during onboarding.
Bring the whole matter into view.
Explore NyayaDesk for your organisation’s records, people and review process.
Talk to us