Microsoft 365 context
Email and Teams context retrieval, calendar reasoning, file discovery, SharePoint knowledge, and Planner operations.
One context. Across your work.
An assistant that connects workplace context to business systems through constrained tools and explicit approval.
Work is distributed across messages, meetings, tasks, files, and business records. An isolated chatbot cannot reliably reason over that environment or safely turn its conclusions into changes.
Connect Microsoft 365 context with Business Central and Dataverse capabilities, using the sequence read, understand, reason, propose, approve, then write.
Select a stage to explore its role in the architecture.
Retrieve scoped information from Outlook, Teams, Calendar, Planner, SharePoint, OneDrive, and Microsoft 365 search.
Resolve cross-source context, calendar constraints, task information, and relevant business records.
Prepare a concrete action or structured artifact such as a quotation comparison.
Present consequential changes for explicit approval before they are applied.
Execute permitted operations through restricted MCP tools and capability allowlists.
Email and Teams context retrieval, calendar reasoning, file discovery, SharePoint knowledge, and Planner operations.
Business Central integration through constrained MCP read/create/update capabilities, with destructive operations disabled.
A capability allowlist for safe Dataverse integration and cross-application reasoning.
Retrieve the relevant evidence across applications before generating a proposed change.
An agent should have the capabilities its task needs. Business Central destructive operations remain disabled.
Separate preparing an action from applying it, and require explicit approval for consequential writes.
I designed and integrated an assistant architecture for cross-system reasoning, structured artifacts, and controlled actions. User-scoped context, restricted tools, and approval gates define how the assistant can interact with connected systems.