Service Design Specification

workforceos-aiworkforceanalytics-service documentation Version: 1.0.134

Scope

This document provides a structured architectural overview of the aiWorkforceAnalytics microservice, detailing its configuration, data model, authorization logic, business rules, and API design. It has been automatically generated based on the service definition within Mindbricks, ensuring that the information reflects the source of truth used during code generation and deployment.

The document is intended to serve multiple audiences:

Note for Frontend Developers: While this document is valuable for understanding business logic and data interactions, please refer to the Service API Documentation for endpoint-level specifications and integration details.

Note for Backend Developers: Since the code for this service is automatically generated by Mindbricks, you typically won’t need to implement or modify it manually. However, this document is especially valuable when you’re building other services—whether within Mindbricks or externally—that need to interact with or depend on this service. It provides a clear reference to the service’s data contracts, business rules, and API structure, helping ensure compatibility and correct integration.

AiWorkforceAnalytics Service Settings

Microservice for computing and delivering AI-powered workforce analytics (e.g., shift optimization, absenteeism insights, staffing trends) for eligible (subscribed) companies. Handles insight calculation, storage, and secure distribution to authorized users. Publishes insight alerts/events for notifications. Strictly enforces company data isolation and subscription gating on all features.

Service Overview

This service is configured to listen for HTTP requests on port 3006, serving both the main API interface and default administrative endpoints.

The following routes are available by default:

The service uses a PostgreSQL database for data storage, with the database name set to workforceos-aiworkforceanalytics-service.

This service is accessible via the following environment-specific URLs:

Authentication & Security

This service requires user authentication for access. It supports both JWT and RSA-based authentication mechanisms, ensuring secure user sessions and data integrity. If a crud route also is configured to require login, it will check a valid JWT token in the request query/header/bearer/cookie. If the token is valid, it will extract the user information from the token and make the fetched session data available in the request context.

Service Data Objects

The service uses a PostgreSQL database for data storage, with the database name set to workforceos-aiworkforceanalytics-service.

Data deletion is managed using a soft delete strategy. Instead of removing records from the database, they are flagged as inactive by setting the isActive field to false.

Object Name Description Public Access Tenant Level
aiInsight Represents an AI-generated insight, analytic, or tip for a company (and optionally individual user) for a specific period/type. Used for staffing recommendations, absenteeism patterns, productivity, and targeted advice. Stores computed detail as JSON and current delivery/processing status. accessPrivate Yes

aiInsight Data Object

Object Overview

Description: Represents an AI-generated insight, analytic, or tip for a company (and optionally individual user) for a specific period/type. Used for staffing recommendations, absenteeism patterns, productivity, and targeted advice. Stores computed detail as JSON and current delivery/processing status.

This object represents a core data structure within the service and acts as the blueprint for database interaction, API generation, and business logic enforcement. It is defined using the ObjectSettings pattern, which governs its behavior, access control, caching strategy, and integration points with other systems such as Stripe and Redis.

Core Configuration

Properties Schema

Display Label Property: insightType — This property is the default display label for records of this data object. Relation dropdowns and record references in the frontend will show the value of this property as the human-readable label.

Property Type Required Description
insightType String Yes Type/classification of the insight (e.g. 'staffingPrediction', 'absenteeismPattern', 'shiftAnomaly', 'productivityTip').
audienceUserId ID No If present, insight is for a specific employee; otherwise company/manager-level.
applicablePeriod Object Yes Date range period to which the insight applies (object: {startDate, endDate})
details Text Yes JSON-stringified blob containing the AI result, message, chart data, or recommendation.
aiStatus Enum Yes Status of the insight; pending (to deliver), delivered (published), error (AI-failed/delivery failed).
deliveredTime Date No Actual delivery/publication time (set when status becomes 'delivered'); null when pending.
companyId ID Yes An ID value to represent the tenant id of the company

Default Values

Default values are automatically assigned to properties when a new object is created, if no value is provided in the request body. Since default values are applied on db level, they should be literal values, not expressions.If you want to use expressions, you can use transposed parameters in any business API to set default values dynamically.

Constant Properties

companyId

Constant properties are defined to be immutable after creation, meaning they cannot be updated or changed once set. They are typically used for properties that should remain constant throughout the object’s lifecycle. A property is set to be constant if the Allow Update option is set to false.

Auto Update Properties

insightType audienceUserId applicablePeriod details aiStatus deliveredTime

An update crud API created with the option Auto Params enabled will automatically update these properties with the provided values in the request body. If you want to update any property in your own business logic not by user input, you can set the Allow Auto Update option to false. These properties will be added to the update API’s body parameters and can be updated by the user if any value is provided in the request body.

Enum Properties

Enum properties are defined with a set of allowed values, ensuring that only valid options can be assigned to them. The enum options value will be stored as strings in the database, but when a data object is created an addtional property with the same name plus an idx suffix will be created, which will hold the index of the selected enum option. You can use the index property to sort by the enum value or when your enum options represent a sequence of values.

Elastic Search Indexing

insightType audienceUserId applicablePeriod details aiStatus deliveredTime companyId

Properties that are indexed in Elastic Search will be searchable via the Elastic Search API. While all properties are stored in the elastic search index of the data object, only those marked for Elastic Search indexing will be available for search queries.

Database Indexing

insightType audienceUserId applicablePeriod aiStatus companyId

Properties that are indexed in the database will be optimized for query performance, allowing for faster data retrieval. Make a property indexed in the database if you want to use it frequently in query filters or sorting.

Secondary Key Properties

companyId

Secondary key properties are used to create an additional indexed identifiers for the data object, allowing for alternative access patterns. Different than normal indexed properties, secondary keys will act as primary keys and Mindbricks will provide automatic secondary key db utility functions to access the data object by the secondary key.

Relation Properties

audienceUserId

Mindbricks supports relations between data objects, allowing you to define how objects are linked together. You can define relations in the data object properties, which will be used to create foreign key constraints in the database. For complex joins operations, Mindbricks supportsa BFF pattern, where you can view dynamic and static views based on Elastic Search Indexes. Use db level relations for simple one-to-one or one-to-many relationships, and use BFF views for complex joins that require multiple data objects to be joined together.

The target object is a sibling object, meaning that the relation is a many-to-one or one-to-one relationship from this object to the target.

On Delete: Set Null Required: No

Filter Properties

insightType audienceUserId applicablePeriod aiStatus companyId

Filter properties are used to define parameters that can be used in query filters, allowing for dynamic data retrieval based on user input or predefined criteria. These properties are automatically mapped as API parameters in the listing API’s that have “Auto Params” enabled.

Business Logic

aiWorkforceAnalytics has got 7 Business APIs to manage its internal and crud logic. For the details of each business API refer to its chapter.

AI Agents

aiWorkforceAnalytics has 1 AI Agent configured. Each agent encapsulates a model configuration, system prompt, input/output pipeline, and optional tool access. Agents support multiple execution modes (task, chat, orchestrator) and modalities (text, image, audio, video, vision).

Agent Name Execution Mode Modality Provider Auth Required
insightGenerator task text openai Yes

For detailed documentation on each agent, refer to:

Edge Controllers

generateInsightHandler

Configuration:

REST Settings:

generateAiInsightStreamHandler

Configuration:

REST Settings:

Service Library

Functions

checkCompanyAIAccess.js

module.exports = function checkCompanyAIAccess(companySubscription) {
  if (!companySubscription) return false;
  if (companySubscription.status !== 'active') return false;
  if (companySubscription.paymentStatus && companySubscription.paymentStatus !== 'paid') return false;
  if (!companySubscription.subscribedFeatures || !Array.isArray(companySubscription.subscribedFeatures)) return false;
  if (companySubscription.expiryDate && new Date(companySubscription.expiryDate) < new Date()) return false;
  return companySubscription.subscribedFeatures.includes('aiInsights') || companySubscription.subscribedFeatures.includes('aiWorkforceAnalytics');
};

generateInsightStream.js

const common = require("common");
const serviceCommon = require("serviceCommon");

module.exports = async (context) => {
  const { goal, period, inputData, insightType, applicablePeriod, details, session } = context;
  
  // Validate required fields
  const requiredFields = ['insightType', 'applicablePeriod', 'details'];
  const missingFields = requiredFields.filter(field => {
    const value = context[field];
    return value === undefined || value === null || value === '';
  });
  
  if (missingFields.length > 0) {
    const error = new Error(`Missing required fields: ${missingFields.join(', ')}`);
    error.statusCode = 400;
    error.code = 'VALIDATION_ERROR';
    throw error;
  }
  
  // Validate applicablePeriod has startDate and endDate
  if (!applicablePeriod || !applicablePeriod.startDate || !applicablePeriod.endDate) {
    const error = new Error('applicablePeriod must contain startDate and endDate');
    error.statusCode = 400;
    error.code = 'VALIDATION_ERROR';
    throw error;
  }
  
  // Build the prompt for AI agent
  const prompt = JSON.stringify({
    goal: goal || 'Generate workforce insight',
    period: period || '',
    inputData: inputData || {},
    insightType: insightType,
    applicablePeriod: applicablePeriod
  });
  
  // Call AI Agent via AgentHub
  try {
    const agentHub = serviceCommon.getAgentHubClient();
    const agentResult = await agentHub.callAgent('insightGenerator', {
      message: prompt,
      session: session
    });
    
    // Return the AI insight result for SSE streaming
    return {
      success: true,
      insightType: insightType,
      applicablePeriod: applicablePeriod,
      details: agentResult.response || details,
      aiStatus: 'delivered',
      generatedContent: agentResult.response || ''
    };
  } catch (agentError) {
    console.error('AI Agent call failed:', agentError);
    throw new Error(`AI insight generation failed: ${agentError.message}`);
  }
};

fetchCompanyWorkforceData.js

const axios = require("axios");
const common = require("common");

module.exports = async (context) => {
  const { companyId, period } = context;
  if (!companyId) throw new Error("companyId is required");

  const authUrl = process.env.AUTH_SERVICE_URL || "http://auth-api:3011";
  const employeeUrl = process.env.EMPLOYEEPROFILE_API_URL || "http://employeeprofile-api:3011";
  const scheduleUrl = process.env.SCHEDULEMANAGEMENT_API_URL || "http://schedulemanagement-api:3011";
  const attendanceUrl = process.env.ATTENDANCEMANAGEMENT_API_URL || "http://attendancemanagement-api:3011";
  const taskUrl = process.env.TASKMANAGEMENT_API_URL || "http://taskmanagement-api:3011";
  const leaveUrl = process.env.LEAVEMANAGEMENT_API_URL || "http://leavemanagement-api:3011";

  const jwtToken = await common.crypto.createInternalServiceJWT();
  const headers = { Authorization: `Bearer ${jwtToken}` };

  try {
    // Build query params with companyId filter
    const companyQuery = `companyId=${companyId}&pageRowCount=0`;
    
    const [
      departmentsRes,
      usersRes,
      employeeProfilesRes,
      shiftsRes,
      attendanceRes,
      tasksRes,
      leaveRes,
    ] = await Promise.allSettled([
      axios.get(`${authUrl}/v1/usergroups?${companyQuery}`, { headers })
        .catch(() => ({ data: { userGroups: [] } })),
      axios.get(`${authUrl}/v1/users?${companyQuery}`, { headers })
        .catch(() => ({ data: { items: [] } })),
      axios.get(`${employeeUrl}/v1/employeeprofiles?${companyQuery}`, { headers })
        .catch(() => ({ data: { employeeProfiles: [] } })),
      axios.get(`${scheduleUrl}/v1/shifts?${companyQuery}`, { headers })
        .catch(() => ({ data: { shifts: [] } })),
      axios.get(`${attendanceUrl}/v1/attendancerecords?${companyQuery}`, { headers })
        .catch(() => ({ data: { attendanceRecords: [] } })),
      axios.get(`${taskUrl}/v1/taskassignments?${companyQuery}`, { headers })
        .catch(() => ({ data: { taskAssignments: [] } })),
      axios.get(`${leaveUrl}/v1/leaverequests?${companyQuery}`, { headers })
        .catch(() => ({ data: { leaveRequests: [] } })),
    ]);

    const departments = departmentsRes.status === "fulfilled" 
      ? departmentsRes.value.data?.userGroups || [] 
      : [];
    const users = usersRes.status === "fulfilled" 
      ? usersRes.value.data?.items || [] 
      : [];
    const employeeProfiles = employeeProfilesRes.status === "fulfilled" 
      ? employeeProfilesRes.value.data?.employeeProfiles || employeeProfilesRes.value.data?.items || [] 
      : [];
    const shifts = shiftsRes.status === "fulfilled" 
      ? shiftsRes.value.data?.shifts || [] 
      : [];
    const attendance = attendanceRes.status === "fulfilled" 
      ? attendanceRes.value.data?.attendanceRecords || attendanceRes.value.data?.items || [] 
      : [];
    const tasks = tasksRes.status === "fulfilled" 
      ? tasksRes.value.data?.taskAssignments || tasksRes.value.data?.items || [] 
      : [];
    const leaveRequests = leaveRes.status === "fulfilled" 
      ? leaveRes.value.data?.leaveRequests || [] 
      : [];

    const stats = {
      totalEmployees: employeeProfiles.length || users.length,
      totalDepartments: departments.length,
      totalShifts: shifts.length,
      totalAttendanceRecords: attendance.length,
      totalTasks: tasks.length,
      pendingLeaveRequests: leaveRequests.filter((l) => l.status === "pending").length,
      approvedLeaveRequests: leaveRequests.filter((l) => l.status === "approved").length,
      rejectedLeaveRequests: leaveRequests.filter((l) => l.status === "rejected").length,
      lateCheckIns: attendance.filter((a) => a.status === "late").length,
      earlyDepartures: attendance.filter((a) => a.status === "earlyLeave").length,
      absences: attendance.filter((a) => a.status === "absent").length,
      completedTasks: tasks.filter((t) => t.status === "completed").length,
      pendingTasks: tasks.filter((t) => t.status === "pending" || t.status === "active").length,
    };

    return {
      stats,
      departments: departments.map((d) => ({ id: d.id, groupName: d.groupName })),
      employees: employeeProfiles.map((p) => {
        const user = users.find((u) => u.id === p.userId);
        const dept = departments.find((d) => d.id === p.departmentId);
        return {
          id: p.id,
          userId: p.userId,
          fullname: user?.fullname || "Unknown",
          email: user?.email || "N/A",
          department: dept?.groupName || "Unassigned",
          position: p.position || "N/A",
          contractType: p.contractType || "N/A",
        };
      }),
      attendanceSummary: attendance.slice(0, 50),
      recentTasks: tasks.slice(0, 20),
      recentLeaveRequests: leaveRequests.slice(0, 20),
    };
  } catch (error) {
    console.error("fetchCompanyWorkforceData error:", error.message);
    return { error: error.message, stats: {}, departments: [], employees: [] };
  }
};

fetchCompanyData.js

const axios = require('axios');

module.exports = async (context) => {
  const { token, baseUrl, userId, companyCodename, companyId } = context;
  
  if (!token || !baseUrl) {
    throw new Error('Token and baseUrl are required');
  }

  const headers = {
    'Authorization': `Bearer ${token}`,
    'Content-Type': 'application/json'
  };

  if (companyCodename) {
    headers['mbx-company-codename'] = companyCodename;
  }

  try {
    // Get current user info
    const userResponse = await axios.get(`${baseUrl}/auth-api/currentuser`, { headers });
    const currentUser = userResponse.data;
    const userCompanyId = companyId || currentUser?.company?.companyId;
    
    // Get employee profiles for the company (use companyId filter if API supports it)
    let employeeCount = 0;
    try {
      const employeeResponse = await axios.get(`${baseUrl}/employeeprofile-api/v1/employeeprofiles?pageNumber=0`, { headers });
      // Filter by company if the API returns all - items should already be filtered by session
      const items = employeeResponse.data?.data || employeeResponse.data?.items || [];
      employeeCount = employeeResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('Employee API error:', e.message);
    }
    
    // Get departments (userGroups)
    let departmentCount = 0;
    try {
      const deptResponse = await axios.get(`${baseUrl}/auth-api/v1/usergroups?pageNumber=0`, { headers });
      const items = deptResponse.data?.data || deptResponse.data?.userGroups || [];
      departmentCount = deptResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('Department API error:', e.message);
    }
    
    // Get today's shifts
    let todayShifts = 0;
    try {
      const today = new Date().toISOString().split('T')[0];
      const shiftsResponse = await axios.get(`${baseUrl}/schedulemanagement-api/v1/shifts?shiftDate=${today}&pageNumber=0`, { headers });
      const items = shiftsResponse.data?.data || shiftsResponse.data?.shifts || [];
      todayShifts = shiftsResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('Today shifts API error:', e.message);
    }
    
    // Get all shifts
    let totalShifts = 0;
    try {
      const allShiftsResponse = await axios.get(`${baseUrl}/schedulemanagement-api/v1/shifts?pageNumber=0`, { headers });
      const items = allShiftsResponse.data?.data || allShiftsResponse.data?.shifts || [];
      totalShifts = allShiftsResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('All shifts API error:', e.message);
    }
    
    // Get task assignments - THIS WAS THE BUG - using wrong endpoint
    let taskCount = 0;
    try {
      const tasksResponse = await axios.get(`${baseUrl}/taskmanagement-api/v1/taskassignments?pageNumber=0`, { headers });
      const items = tasksResponse.data?.data || tasksResponse.data?.items || [];
      taskCount = tasksResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('Tasks API error:', e.message);
    }
    
    // Get individual tasks for this user
    let myTaskCount = 0;
    try {
      const myTasksResponse = await axios.get(`${baseUrl}/taskmanagement-api/v1/myindividualtasks?pageNumber=0`, { headers });
      const items = myTasksResponse.data?.data || myTasksResponse.data?.items || [];
      myTaskCount = myTasksResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('My tasks API error:', e.message);
    }
    
    // Get attendance records
    let attendanceCount = 0;
    try {
      const attendanceResponse = await axios.get(`${baseUrl}/attendancemanagement-api/v1/attendancerecords?pageNumber=0`, { headers });
      const items = attendanceResponse.data?.data || attendanceResponse.data?.items || [];
      attendanceCount = attendanceResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('Attendance API error:', e.message);
    }
    
    // Get leave requests
    let leaveCount = 0;
    try {
      const leaveResponse = await axios.get(`${baseUrl}/leavemanagement-api/v1/leaverequests?pageNumber=0`, { headers });
      const items = leaveResponse.data?.data || leaveResponse.data?.leaveRequests || [];
      leaveCount = leaveResponse.data.rowCount || items.length;
    } catch (e) {
      console.log('Leave API error:', e.message);
    }

    return {
      user: currentUser,
      employeeCount,
      departmentCount,
      todayShifts,
      totalShifts,
      taskCount,
      myTaskCount,
      attendanceCount,
      leaveCount,
      today: new Date().toISOString().split('T')[0],
      companyCodename: companyCodename || currentUser?.company?.codename || 'unknown',
      userCompanyId: userCompanyId
    };
  } catch (error) {
    console.error('Error fetching company data:', error.message);
    if (error.response) {
      console.error('Response status:', error.response.status);
      console.error('Response data:', error.response.data);
    }
    throw new Error(`Failed to fetch company data: ${error.message}`);
  }
};

Edge Functions

generateAiInsightStreamHandler.js

const axios = require("axios");

module.exports = async (req) => {
  const baseUrl = "https://workforceos.prw.mindbricks.com";
  const authHeader = req.headers?.authorization;
  const token = authHeader?.substring(7);
  const companyCodename = req.headers["mbx-company-codename"];

  if (!token)
    return {
      status: 401,
      headers: { "Content-Type": "text/event-stream" },
      content: `event: error\ndata: {"error":"Authentication required"}\n\n`,
    };

  const goal = req.body?.goal || req.body?.query;
  if (!goal)
    return {
      status: 400,
      headers: { "Content-Type": "text/event-stream" },
      content: `event: error\ndata: {"error":"Goal or query is required"}\n\n`,
    };

  const headers = {
    Authorization: `Bearer ${token}`,
    "Content-Type": "application/json",
  };
  if (companyCodename) headers["mbx-company-codename"] = companyCodename;

  try {
    // Call the workforceAssistant AI agent in agentHub
    const agentUrl = `${baseUrl}/agenthub-api/agents/workforceAssistant`;
    
    const agentRes = await axios.post(
      agentUrl,
      { message: goal },
      { headers }
    );

    const aiResponse = agentRes.data;

    // Stream the response
    const sse = 
      `event: start\ndata: {}\n\n` +
      `event: chunk\ndata: {"chunk": "Workforce AI Analysis", "type": "title"}\n\n` +
      `event: chunk\ndata: {"chunk": ${JSON.stringify(aiResponse)}, "type": "summary"}\n\n` +
      `event: complete\ndata: {"status": "completed", "insightType": "companySpecific"}\n\n`;

    return {
      status: 200,
      headers: { "Content-Type": "text/event-stream" },
      content: sse,
    };
  } catch (agentError) {
    console.error("AI Agent call failed:", agentError.message);
    
    // Fallback response if AI agent fails
    const sse = 
      `event: start\ndata: {}\n\n` +
      `event: chunk\ndata: {"chunk": "Workforce AI", "type": "title"}\n\n` +
      `event: chunk\ndata: {"chunk": "I'm having trouble accessing the AI service right now. Please try again in a moment.", "type": "summary"}\n\n` +
      `event: complete\ndata: {"status": "completed", "insightType": "general"}\n\n`;

    return {
      status: 200,
      headers: { "Content-Type": "text/event-stream" },
      content: sse,
    };
  }
};

streamAiInsight.js

// Wrapper for generateAiInsightStreamHandler with login enforcement
const handler = require('./generateAiInsightStreamHandler');
module.exports = async (request) => {
  return handler(request);
};

testFunction.js

module.exports = async (req) => { return { status: 200, content: 'test' }; };

generateInsightHandler.js

const axios = require('axios');

module.exports = async (req, res) => {
  try {
    // Get token from request
    const authHeader = req.headers['authorization'] || req.headers['Authorization'];
    const token = authHeader ? authHeader.replace('Bearer ', '') : null;
    
    if (!token) {
      return res.status(401).json({ 
        status: 'ERR', 
        message: 'Authorization token required' 
      });
    }

    // Get company codename from header
    const companyCodename = req.headers['mbx-company-codename'] || req.headers['mbx-Company-Codename'];
    
    // Build base URL from request
    const protocol = req.headers['x-forwarded-proto'] || 'https';
    const host = req.headers['host'];
    const baseUrl = `${protocol}://${host}`;

    const headers = {
      'Authorization': `Bearer ${token}`,
      'Content-Type': 'application/json'
    };

    if (companyCodename) {
      headers['mbx-company-codename'] = companyCodename;
    }

    // Get current user info
    const userResponse = await axios.get(`${baseUrl}/auth-api/currentuser`, { headers });
    const currentUser = userResponse.data;
    
    if (!currentUser || !currentUser.userId) {
      return res.status(401).json({ 
        status: 'ERR', 
        message: 'Invalid token or user not found' 
      });
    }

    // Fetch all company data using the APIs
    const today = new Date().toISOString().split('T')[0];
    
    // Get employee profiles for the company
    let employeeCount = 0;
    try {
      const employeeResponse = await axios.get(`${baseUrl}/employeeprofile-api/v1/employeeprofiles?pageNumber=0`, { headers });
      employeeCount = employeeResponse.data.rowCount || 0;
    } catch (e) {
      console.log('Employee API error:', e.message);
    }
    
    // Get departments (userGroups)
    let departmentCount = 0;
    try {
      const deptResponse = await axios.get(`${baseUrl}/auth-api/v1/usergroups?pageNumber=0`, { headers });
      departmentCount = deptResponse.data.rowCount || 0;
    } catch (e) {
      console.log('Department API error:', e.message);
    }
    
    // Get today's shifts
    let todayShifts = 0;
    try {
      const shiftsResponse = await axios.get(`${baseUrl}/schedulemanagement-api/v1/shifts?shiftDate=${today}&pageNumber=0`, { headers });
      todayShifts = shiftsResponse.data.rowCount || 0;
    } catch (e) {
      console.log('Today shifts API error:', e.message);
    }
    
    // Get all shifts
    let totalShifts = 0;
    try {
      const allShiftsResponse = await axios.get(`${baseUrl}/schedulemanagement-api/v1/shifts?pageNumber=0`, { headers });
      totalShifts = allShiftsResponse.data.rowCount || 0;
    } catch (e) {
      console.log('All shifts API error:', e.message);
    }
    
    // Get task assignments
    let taskCount = 0;
    try {
      const tasksResponse = await axios.get(`${baseUrl}/taskmanagement-api/v1/taskassignments?pageNumber=0`, { headers });
      taskCount = tasksResponse.data.rowCount || 0;
    } catch (e) {
      console.log('Tasks API error:', e.message);
    }
    
    // Get individual tasks for this user
    let myTaskCount = 0;
    try {
      const myTasksResponse = await axios.get(`${baseUrl}/taskmanagement-api/v1/myindividualtasks?pageNumber=0`, { headers });
      myTaskCount = myTasksResponse.data.rowCount || 0;
    } catch (e) {
      console.log('My tasks API error:', e.message);
    }
    
    // Get attendance records
    let attendanceCount = 0;
    try {
      const attendanceResponse = await axios.get(`${baseUrl}/attendancemanagement-api/v1/attendancerecords?pageNumber=0`, { headers });
      attendanceCount = attendanceResponse.data.rowCount || 0;
    } catch (e) {
      console.log('Attendance API error:', e.message);
    }
    
    // Get leave requests
    let leaveCount = 0;
    try {
      const leaveResponse = await axios.get(`${baseUrl}/leavemanagement-api/v1/leaverequests?pageNumber=0`, { headers });
      leaveCount = leaveResponse.data.rowCount || 0;
    } catch (e) {
      console.log('Leave API error:', e.message);
    }

    // Build company data summary
    const companyData = {
      employeeCount,
      departmentCount,
      todayShifts,
      totalShifts,
      taskCount,
      myTaskCount,
      attendanceCount,
      leaveCount,
      today,
      companyCodename: companyCodename || currentUser?.company?.codename || 'unknown',
      userCompanyId: currentUser?.company?.companyId || currentUser?.companyId
    };

    // Get user prompt from request
    const userPrompt = req.body?.prompt || 'Provide a workforce overview';
    
    // Build enriched prompt with company data
    const enrichedPrompt = `ACTUAL COMPANY DATA (from database):
- Employee Count: ${companyData.employeeCount}
- Department Count: ${companyData.departmentCount}
- Shifts Today (${today}): ${companyData.todayShifts}
- Total Shifts: ${companyData.totalShifts}
- Task Assignments: ${companyData.taskCount}
- My Tasks: ${companyData.myTaskCount}
- Attendance Records: ${companyData.attendanceCount}
- Leave Requests: ${companyData.leaveCount}
- Company Codename: ${companyData.companyCodename}

USER QUESTION: ${userPrompt}

IMPORTANT: Use ONLY the numbers above. Do not make up or hallucinate any data. If a count is 0, report it as 0.

Provide your response as a JSON object with this exact structure:
{
  "title": "Brief insight title",
  "insightType": "companySpecific",
  "summary": "2-3 sentence answer using the actual numbers above",
  "data": {
    "keyMetrics": [{"metric": "Name", "value": number}],
    "findings": ["Finding 1"],
    "recommendations": []
  },
  "audienceType": "company",
  "suggestion": "Optional suggestion"
}`;

    // Call the AI agent
    const aiResponse = await axios.post(
      `${baseUrl}/aiworkforceanalytics-api/v1/aiinsight`,
      {
        prompt: enrichedPrompt
      },
      { headers }
    );

    // Return the insight
    res.json({
      status: 'OK',
      insight: aiResponse.data.insight
    });

  } catch (error) {
    console.error('Edge controller error:', error.message);
    if (error.response) {
      console.error('Response status:', error.response.status);
      console.error('Response data:', error.response.data);
    }
    
    if (!res.headersSent) {
      res.status(500).json({
        status: 'ERR',
        message: error.message || 'Internal server error',
        error: error.response?.data || error.message
      });
    }
  }
};

This document was generated from the service architecture definition and should be kept in sync with implementation changes.