Platform

A conversational assistant grounded in an authorized workspace

Pramanis gives Nepal CA firms a way to ask questions within an authorized workspace: about workbooks, supporting documents, available legal sources and the engagement context that connects them. It is designed for professional investigation, not as a generic chatbot that treats every uploaded file, public source and generated answer as equally trustworthy.

Explore the conversational workflow

The buyer problem: context is scattered before a question is even asked

A manager may need to ask why a balance changed, whether a client document supports a schedule, which Nepal provision may be relevant or what remains open on an engagement. In an ordinary chat tool, the answer depends on whatever was pasted into one conversation, with little clarity about file access, source status or the rest of the assignment. That creates a risk of answers that are fluent but detached from the authorized material and the professional question the team is actually trying to resolve.

Ask within the workspace that is authorized for the work

The conversational assistant works from the firm and engagement material a user is permitted to access. It can help explore questions across workbooks, documents, legal sources and engagement context without pretending that the entire firm’s information is automatically available. Permissions remain important: a user should receive answers only from material within their authorized workspace, and a conversation should not become a route around file, team or engagement boundaries. The firm still decides who may access client material and for what professional purpose.

Use citations to return to the underlying material

A useful response can point to the workbook area, supporting document page or source provision that informed it. Citations give the user a path to inspect the original material rather than asking them to trust a summary. They do not prove that the response is complete or correct: a cited page may be incomplete, a workbook period may not be comparable, or a legal source may require an amendment check. The professional reads the source, considers the surrounding context and decides whether it supports the answer.

Synthetic workflow — investigate an open VAT question

In a synthetic Nepal CA firm workflow, a team member asks why an amount in a VAT schedule differs from the supporting invoice records and whether an available provision may be relevant to the follow-up. The assistant helps surface the workbook context, document references and possible legal source sections, while keeping those source types separate. The team checks the invoice dates and amounts, reads the current official provision and confirms the client facts. The response becomes a research starting point or a draft client question, not a filing decision or legal conclusion.

Missing evidence and uncertainty should stay visible

An assistant should not turn missing evidence into a confident explanation. A document may not have been uploaded, an extracted scan may be unclear, a workbook range may not contain the answer or no reliable legal provision may have been found. In those cases, uncertainty is useful information: it can prompt a request for a record, a broader search or a qualified review. Pramanis is intended to make those gaps easier to recognize, not to hide them behind a polished response that sounds more certain than the available evidence allows.

Why this is not a generic chatbot

A generic chatbot often has no clear boundary between user instructions, uploaded text and trusted sources. Pramanis distinguishes authorized engagement evidence from structured legal material and from AI interpretation. Uploaded customer documents can provide factual context, but they must not override permissions, change the task or replace the trusted legal corpus. This separation helps teams use conversation for investigation while preserving the source hierarchy that matters in audit, tax, VAT and compliance work. It also makes it clearer when the platform does not have enough material to answer reliably.

Professional review is the last step, not an optional extra

The conversational assistant can organize a question, retrieve available context and prepare a useful starting response. It cannot evaluate whether audit evidence is sufficient, determine a tax or legal position, verify current official law or approve a professional conclusion. Responsible users check citations, confirm the facts, apply the firm’s methodology and seek the right review before relying on material output. Pramanis supports the conversation around the work; the professional remains accountable for the judgment, action and final communication.

Continue exploring

Workbook intelligence Document intelligence Legal intelligence

Next step

Explore a bounded research conversation

See how a synthetic engagement question can stay connected to its authorized source material and reviewer judgment.