project-60f26edc

Here is a master prompt you can use to generate the AI web app. It is designed around the **sales rep handover** wedge, with scoped permissions, knowledge capture, and a simple enterprise UI.[1][2][3][4] ## Master prompt You are an expert product engineer, UX architect, and full-stack AI app builder. Build a secure, multi-tenant **AI Handover Web App** for enterprises that captures the knowledge of departing employees and turns it into an interactive handover assistant for managers and replacements. ### Product goal The app helps a manager answer: “What did this person know, do, and usually handle?” It must focus on one narrow use case first: **sales rep departure handover**. The app should ingest CRM data, email metadata/content, calendar events, documents, notes, and a short exit interview, then generate a concise knowledge-transfer brief and an ask-anything interface for the manager. The app should not try to become a full AI operating system. It should be a scoped, permission-aware enterprise product, similar in spirit to how Microsoft and Google manage AI agents with admin controls, user scoping, and data-source governance.[2][3][4][5][1] ### Core users - Sales manager. - Departing employee. - New owner / replacement rep. - Org admin / security admin. ### Core workflow 1. Admin connects approved data sources. 2. Manager selects a departing employee. 3. System imports data from CRM, email, calendar, and docs. 4. App runs a guided exit interview to capture tacit knowledge. 5. App produces: - executive summary, - account-by-account handover, - deal status and risks, - recurring patterns and playbooks, - key contacts and relationships, - open questions and missing information. 6. Manager can ask natural-language questions over the handover context. 7. Access is restricted by role, scope, and permissions. ### Non-negotiable product rules - Keep the first version limited to sales handover. - Do not build CEO dashboards, org-wide priority engines, cross-department planning, or generic automation. - Do not make the app depend on a huge number of integrations. - Use permission-aware retrieval only. - Never expose data across users, teams, or organizations without explicit access controls. - Make the app auditable, explainable, and admin-governed. This matches current enterprise AI control patterns.[3][4][1][2] ### Functional requirements - Authentication with org-based tenant separation. - Role-based access control for admin, manager, employee, and viewer. - Data connector setup for CRM, email, calendar, and documents. - Employee offboarding checklist. - AI-generated handover brief. - Q&A assistant grounded in imported data. - Change log and audit trail for all generated outputs. - Manual review and edit before publishing the final handover. - Export to PDF and share internally with permissions. ### AI behavior - Summarize only from authorized sources. - Cite which source types were used in each answer. - Distinguish between facts, inferred insights, and missing data. - Ask clarifying questions when important context is absent. - Highlight deal risk, customer relationships, active commitments, and unanswered tasks. - Never hallucinate account ownership, contract status, or next steps. ### Data model Design entities for: - Organization. - User. - Role. - Employee profile. - Departure event. - CRM account. - Deal. - Contact. - Email thread. - Calendar event. - Document. - Exit interview response. - Handover brief. - Question and answer. - Audit log. ### UI requirements Create a clean enterprise dashboard with: - left sidebar navigation, - employee selector, - source status panel, - handover brief editor, - ask-anything chat panel, - risk flags, - export/share actions, - admin controls for connectors and permissions. ### Security and governance - Tenant isolation. - Source-level access controls. - User-level visibility rules. - Admin approval for connectors. - Audit logging. - Data minimization. - Ability to disable a source without breaking the app. - Clear indicator of what data the AI can and cannot use. Enterprise AI systems increasingly rely on exactly these kinds of admin-scoped source controls and agent governance.[4][5][1][2][3] ### MVP acceptance criteria - A manager can upload or sync a rep’s data. - The app generates a useful handover brief in under 2 minutes. - The manager can ask at least 10 meaningful questions and get grounded answers. - The system respects permissions. - The output is immediately useful for a replacement rep within one day. ### Output format Return: 1. Product architecture. 2. Database schema. 3. Backend API design. 4. Frontend page structure. 5. AI prompt strategy. 6. Security model. 7. MVP scope. 8. Implementation plan for the first 30 days. ### Tone and design principles - Enterprise-ready. - Simple. - Trustworthy. - Scoped. - Not flashy. - Built for real adoption, not demo theater. ### Final instruction If there are multiple reasonable design choices, choose the simplest secure option that helps ship the MVP fastest. Prioritize working software over abstract platform thinking. ## Optional shorter version Build a permission-aware enterprise web app for sales rep handover. Ingest CRM, email, calendar, docs, and exit interview notes for a departing employee. Generate a manager-facing handover brief, account summary, risk flags, and an ask-anything assistant grounded only in approved sources. Include tenant isolation, role-based access control, audit logs, admin-managed connectors, and manual review before publishing. Do not build a full AI OS; build a narrow, useful knowledge-transfer product first.[1][2][3][4] If you want, I can turn this into a **code-generation prompt for Next.js + Supabase + OpenAI** next. Sources [1] Manage Copilot agents in the Microsoft 365 admin center https://learn.microsoft.com/en-us/microsoft-365/admin/manage/manage-copilot-agents-integrated-apps?view=o365-worldwide [2] Manage agents for Microsoft 365 Copilot | Microsoft Learn https://learn.microsoft.com/en-us/microsoft-365/copilot/extensibility/manage [3] Copilot Control System Management Controls | Microsoft Learn https://learn.microsoft.com/en-us/microsoft-365/copilot/copilot-control-system/management-controls [4] Microsoft 365 Copilot agents governance visual guide https://learn.microsoft.com/en-us/copilot/microsoft-365/agent-essentials/m365-agents-visual-map [5] Introducing Workspace Intelligence, with admin controls https://workspaceupdates.googleblog.com/2026/04/introducing-workspace-intelligence-with-admin-controls.html [6] [PDF] Administering and Governing Agents - Microsoft 365 Adoption https://adoption.microsoft.com/files/copilot-studio/Agent-governance-whitepaper.pdf [7] Controlar a Inteligência do Workspace para recursos de IA ... https://knowledge.workspace.google.com/admin/generative-ai/workspace-intelligence/control-workspace-intelligence?hl=pt-br [8] Control Agents as a Microsoft 365 Admin | Copilot Control System updates https://www.youtube.com/watch?v=cQU1GTm14S8 [9] What governance, access, and action controls can we apply to a ... https://learn.microsoft.com/en-us/answers/questions/5852075/what-governance-access-and-action-controls-can-we [10] Governance 101 for Microsoft Copilot Agents https://www.youtube.com/watch?v=vygjaGHsvoE