Ask Sanad in natural language: “Where's the supplier contract expiring next month?” or “Build an approval path for new contracts” or “Archive the closed project's documents.” Sanad understands the context of your system and documents, and shows the answer or prepares the action and runs it after you confirm.
Instead of moving between screens and folders, you start straight from your goal.
Sanad is a conversational AI agent inside the Mstnd system. It reads the system data and the document content the user is allowed to access, answers questions, searches by meaning, fills metadata fields, and helps carry out actions such as building workflows, archiving, sending circulars, and reviewing tasks and document statuses.
A manager needs to know which contracts will expire within the month and the overdue tasks tied to them, then create a renewal path. The information exists, but reaching it requires moving between search, the calendar, tasks, and workflows and carrying out several steps manually.
The manager talks to Sanad in natural language. The agent searches within the documents and data allowed to them, gathers the answer from the system context, then helps them fill fields, build the path, or carry out the requested action without having to know where every function is in the interface.
Turning an administrative question into an answer and an action within the same context, while respecting the user's permissions.
The system may hold everything you need, but reaching the answer requires opening multiple reports, filtering lists, then opening a document and reading its pages. And after finding the information, another chain of clicks begins to carry out the action.
This burden keeps the knowledge trapped in the interface. A new employee doesn't know where to look, the manager asks for a manual report, and a simple action requires prior knowledge of where every option is.
Ask about users, departments, permissions, tasks, correspondence, document statuses, and content size. Sanad can read the operational data and show a direct answer like “How many overdue tasks does the legal department have?” or “Which contracts are awaiting approval?”
Sanad isn't limited to file names. It reads the text, extracted data, and metadata fields, and can answer a question inside a contract, invoice, or letter, and show the document it relied on.
On three levels: inside a document, ask about clauses, dates, parties, and amounts, or request filling a metadata form; inside a folder, analyze the set of files and subdocuments and extract a combined picture of their content; and at the system level, ask about tasks, correspondence, statuses, departments, and documents within your permissions.
Define the fields you need, such as contract number, party, start and end dates, value, and category, and Sanad reads the document and fills them with high accuracy. The employee can review the values, then approve them so they become searchable, analyzable data for reports.
Say: “I want new contracts to go first to Legal, then Finance, then the manager's signature, with each step taking 2 days.” Sanad turns the description into a structured path and shows the steps, owners, and deadlines before it's approved.
Sanad helps with:
When an action is requested, Sanad understands the goal and asks for any missing information, then shows a summary of what will change. After the user confirms, it carries out the operation with the user's own permissions and records that it was done by Sanad on their behalf. The agent doesn't grant a permission that doesn't already exist, and doesn't reveal a document the user can't open.
A legal affairs manager types: “Show the contracts expiring within 90 days that have no renewal task started.” Sanad gathers the results from the documents and tasks and shows a list linked to its sources. The manager then asks: “Create a renewal task for each contract and assign it to the contracts officer 60 days before expiry.” Sanad shows the preview, and after confirmation it creates the tasks and records the actions.
Yes. It uses the analyzed content, metadata, and semantic search to answer questions and show the documents closest to what's intended.
It works on all three levels: inside a document, inside a folder, and at the system level and its operational data, within the user's permissions.
Yes. You can describe the steps, actions, and owners in natural language, and Sanad turns them into a structured workflow that it shows you before it's approved.
Sanad shows a summary of the action and its impact, then carries it out after the user confirms, recording the operation in the audit log and linking it to the requester's identity.
Yes. A metadata form can be defined for contracts, invoices, letters, or others, and Sanad extracts the appropriate values from the content and fills the defined fields.
No. Sanad helps prepare the document, the path, and the action request, but signing and stamping remain a direct act by the authorized person after their identity is verified.
Don't look for where the action is; tell Sanad what you want to get done.
Book an AI-agent demo in Mstnd