ZenithAIby Data Ingenious Request a Demo

Platform · Workspaces

Workspaces: from your documents to answers you can trust, everywhere.

A ZenithAI workspace is a team’s shelf of documents that answers back: policies, manuals, circulars, contracts, price lists, past reports. You fill it once, keep it current, decide who may use it, and then ask it in plain language, in chat or from your own software. Every answer comes from those documents, names its sources, and says plainly when the documents don’t cover the question. It all runs on your organisation’s own servers.

Follow the journey See it on your own documents

The whole picture

Three ways in. Two ways to ask.

Documents reach a workspace from people, from public websites and from your own systems. Once ready, they answer people in chat and applications through the API, under the same access rules.

How a ZenithAI workspace works Three sources, uploads in the app, website pages and files added by your software through the API, flow into a workspace on your own servers. The workspace answers people in chat and your applications through the API. Upload in the appImport website pagesAdd from your software Ask in chatAsk from your apps PDF, Word, Excel, CSVone page, or up to 25POST /workspace/documents whole, or one documentPOST /workspace YOUR SERVERS Workspace documents, ready to answer persona and pinned facts who may use it sources and “not found”
Everything inside the boundary runs on your organisation’s own servers. The web is used only when a person asks for it in chat.

The journey

Seven stages, one governed workspace.

  1. 01

    Create a workspace

    Anyone can create personal workspaces (up to 20 each) for their own work. Give it a name, such as HR policies 2026, and instructions for how it should answer, for example for branch staff, in simple words. Tell it standing facts to remember, a threshold, a convention, a GST rate, and every future chat starts knowing them.

    For a department or the whole organisation, people with the Org workspace manager role create organisation workspaces and decide who may use each one.

  2. 02

    Fill it, three ways

    • Upload in the app. PDF, Word (DOCX), Excel (XLSX), CSV, text and Markdown files, up to 3 MB each and 100 MB per workspace. Organisation workspaces are filled from Organisation Knowledge.
    • Import from a website. One public page, or up to 25 pages from a site, fetched and indexed on your own hardware. Website Knowledge →
    • Add from your own software. A document system, an HR portal or a nightly job can add a TXT, PDF, DOCX, XLSX or CSV file to its owner’s personal workspace through the API, when an administrator allows it for that key. Enterprise Knowledge API →
  3. 03

    Let it get ready

    Each document is read and indexed in the background, and used for answers as soon as it is ready. Scanned PDFs are read with OCR on your server. Spreadsheets are kept as tables, so questions such as totals and averages are calculated, not guessed. A file that can’t be read is rejected straight away, and a document whose indexing fails says so and can simply be added again.

  4. 04

    Decide who may use it

    Personal workspaces are private to the person who made them: another person’s workspace doesn’t just refuse access, it doesn’t appear at all. An administrator can publish a workspace for the whole organisation, which everyone can then ask but only its owner and administrators can change. An organisation workspace is open to everyone in the organisation or only selected people, as its manager chooses, with an activity log.

  5. 05

    Ask in chat

    Start a chat, pick the workspace, and ask the whole workspace or one document in it. Under each answer ZenithAI shows what it looked at, such as From 3 of 3 documents or Searched 4 of 12 documents, with the names.

    “How many casual leaves does a new joiner get in the first year?”

    In a workspace chat, facts come only from the workspace documents. When they don’t hold the answer, ZenithAI says so and names the documents it checked, rather than filling the gap from general knowledge. The web is used only when the person asks for it, and every such search is logged.

  6. 06

    Ask from your applications

    The same workspace answers your intranet, help desk, branch app or team chat bot through the Enterprise Knowledge API. Each answer carries its sources and a found flag that is false when the documents don’t say. An API key acts as the person who made it, so an application sees exactly the workspaces that person can open, and loses them the moment the person does. Asking stores nothing.

  7. 07

    Keep it current, and keep it governed

    Answers are only as current as the documents, so retire a superseded circular when its replacement arrives: delete it in the app and add the new one. The list of questions the workspace couldn’t answer is your list of documents to add next. Administrators see API and chat usage, organisation workspaces keep an activity log, and server logs record ids and counts, not questions, answers or document names.

Where teams use them

One workspace per body of knowledge.

Illustrative patterns. Each workspace holds the documents its team already maintains, and answers the questions that team is asked every day.

WorkspaceFilled fromWho asksA typical question
HR policiesLeave, travel and conduct policiesEvery employee, in chat and the intranet“What is the hotel limit for a manager in a metro city?”
Branch operationsCirculars and process notes, added by the document system every eveningBranch staff“Which documents does a partnership firm need to open a current account?”
Product supportManuals, warranty terms and the product site, importedThe help desk, before replying“What is the warranty on the 1.5 tonne model?”
Academic regulationsRegulations, notices and the fee scheduleStudents and faculty, through the portal“What is the minimum attendance to sit the semester exam?”
ContractsSigned agreements, added as they are signedLegal and procurement, one contract at a time“Does this agreement allow termination without notice?”
Monthly reportingThe month’s sales CSVManagers“What were total sales in the West region?” (calculated from the file)

Why people trust the answers

Grounded, sourced, and honest about gaps.

A workspace answer is built to be checked. It comes from your documents, names them, calculates numbers from the data, and says “not in the documents” instead of guessing.

  • 01Only your documents · facts in a workspace chat come from the workspace, not from general knowledge or an uninvited web search
  • 02Sources with every answer · the documents behind it, so anyone can open the policy and check
  • 03“Not found” is an answer · in chat it is said plainly; through the API it is a found flag your software can read
  • 04Access follows the person · in chat and through the API alike, nobody sees a workspace they couldn’t open in the app
  • 05On your own servers · documents, indexing and answers stay inside your infrastructure. Security & governance →

Wrong answers are reduced, not ruled out. Show the sources and let people check anything that matters. ZenithAI’s architecture supports data-residency and data-minimisation objectives, including obligations under India’s DPDP Act; it is not a certification.

Step-by-step guides

Every stage, in the documentation.

StageFor people at workFor developers and administrators
Create and fillWorkspaces: create, add documents, askAPI: add a document and check its status
ShareShare with your teamOrganisations and Org workspace managers
AskChat with ZenithAIAsk your workspace from your app
GovernWeb search in workspace chatsLet a key add documents

Questions

Workspaces, answered.

What is a ZenithAI workspace?
A workspace is a set of documents, such as policies, manuals, circulars and reports, that people and applications can ask questions of in plain language. It also holds instructions for how to answer and standing facts to remember. Answers come from its documents, name their sources and say when the documents don’t cover the question.

How do documents get into a workspace?
Three ways. People upload PDF, Word, Excel, CSV, text and Markdown files in the app, up to 3 MB each and 100 MB per workspace. They can import one public web page or up to 25 pages from a site. And your own software can add files through the API, when an administrator allows it for the key.

Who can see a workspace?
Personal workspaces are private to the person who made them; other people don’t even see they exist. An administrator can publish a workspace for the whole organisation. Organisation workspaces are run by Org workspace managers, who open each one to everyone in the organisation or only to selected people.

Does a workspace answer from outside its documents?
No. In a workspace chat, facts come only from the workspace documents. When they don’t hold the answer, ZenithAI says so and names the documents it checked. The web is used only when the person asks for it, and every such search is logged.

Can our own applications use a workspace?
Yes. Through the Enterprise Knowledge API, an intranet, help desk or chat bot asks a workspace and gets the answer with its sources and a found flag. An API key acts as the person who made it, so it sees only the workspaces that person can open. Asking stores nothing.

Where are workspace documents stored?
On your organisation’s own ZenithAI servers. Reading, OCR for scanned files, indexing and answering all happen there. A document stays in its workspace until someone deletes it, and server logs record ids and counts, not questions, answers or document names.

How do we keep a workspace up to date?
Add new documents as they are approved, and delete superseded versions so they can’t be quoted. A program can do the adding for you through the API. The questions a workspace couldn’t answer show which documents to add next.

Why this matters: Where do your documents go when you ask AI?, One company, one answer: a private internal knowledge base and AI for enterprise knowledge management: what the research says.

Bring a few real documents

Build a workspace with us. Ask it your hardest question.

In a demonstration, put your own policies or manuals in a workspace, share it, and ask it in chat and through the API, including a question the documents can’t answer.

Request an Enterprise Demonstration Read the workspace guide