📊☁️
Agentic AI in Action
Salesforce Reporting, Excel Export & Trend Analytics
One conversation. Four connectors. Authenticated data. Visual insights.
The User Prompt
"Pull all customer support cases from Salesforce for Q1, export to Excel, and show me product trend charts"
One sentence triggers a 10-step agentic workflow — authenticating to Salesforce via OAuth, querying case data, enriching with account details, exporting to Excel with formatting, generating trend charts, and delivering the final report.
What ContextWeaver Does
1
Authenticate to Salesforce
Salesforce API
sf_authenticate(instance="mycompany.salesforce.com")OAuth client credentials from vault:
@vault:cw-org-salesforce-client-secret
"Connected to Salesforce as service account. Access token valid for 2 hours."
OAuth client ID + secret from vault — never exposed to user or code
2
Query Q1 Case Data
Salesforce API
sf_query(soql="SELECT Id, CaseNumber, Subject, Status, Priority, Product__c, CreatedDate, ClosedDate, Account.Name FROM Case WHERE CreatedDate >= 2026-01-01T00:00:00Z AND CreatedDate < 2026-04-01T00:00:00Z")
"Found 847 cases for Q1 2026. Breakdown: 312 open, 498 closed, 37 escalated."
3
Enrich with Account Details
Salesforce API
sf_query(soql="SELECT Id, Name, Industry, AnnualRevenue, BillingCountry FROM Account WHERE Id IN ({account_ids})")
"Enriched 847 cases with account industry, revenue tier, and region data."
4
Analyze Product Trends
AI Analysis
analyze_trends(data=cases, group_by="Product__c", time_field="CreatedDate", interval="week")
"Trend analysis complete. Key findings:
• Platform Core: 34% of cases, trending up 12% week-over-week in March
• Mobile SDK: spike in 'crash on launch' cases starting Feb 15
• API Gateway: 89% resolution rate (highest), avg 2.1 days to close
• Enterprise SSO: 3 escalated cases from Fortune 500 accounts"
• Platform Core: 34% of cases, trending up 12% week-over-week in March
• Mobile SDK: spike in 'crash on launch' cases starting Feb 15
• API Gateway: 89% resolution rate (highest), avg 2.1 days to close
• Enterprise SSO: 3 escalated cases from Fortune 500 accounts"
5
User Asks Follow-Up
Voice Input
"Which customers have the most open critical cases?"
sf_query(soql="...WHERE Priority='Critical' AND Status!='Closed' GROUP BY Account.Name ORDER BY COUNT(Id) DESC LIMIT 10")
"Top 3 accounts with open critical cases:
1. Acme Corp — 7 critical (Enterprise SSO + API Gateway)
2. GlobalTech Inc — 4 critical (Platform Core)
3. MedStar Health — 3 critical (Mobile SDK crashes)"
1. Acme Corp — 7 critical (Enterprise SSO + API Gateway)
2. GlobalTech Inc — 4 critical (Platform Core)
3. MedStar Health — 3 critical (Mobile SDK crashes)"
6
Generate Excel Workbook
Excel / OneDrive
create_excel(sheets=["All Cases", "By Product", "By Priority", "Top Accounts", "Trends"])format_excel(headers=True, auto_filter=True, freeze_panes=True, column_widths="auto")
"Excel workbook created with 5 sheets:
• All Cases: 847 rows with case details, account, product, dates
• By Product: pivot summary with counts, avg resolution time
• By Priority: critical/high/medium/low breakdown
• Top Accounts: top 20 accounts by case volume
• Trends: weekly case counts by product"
• All Cases: 847 rows with case details, account, product, dates
• By Product: pivot summary with counts, avg resolution time
• By Priority: critical/high/medium/low breakdown
• Top Accounts: top 20 accounts by case volume
• Trends: weekly case counts by product"
7
Generate Trend Charts
Excel / OneDrive
add_chart(sheet="Trends", type="line", title="Weekly Case Volume by Product", x="Week", y="Count", series="Product")add_chart(sheet="By Priority", type="bar", title="Q1 Cases by Priority")add_chart(sheet="By Product", type="pie", title="Case Distribution by Product")
"Added 3 charts: weekly trend line, priority bar chart, and product distribution pie chart."
8
Upload to OneDrive / SharePoint
Microsoft Graph
upload_file(path="/Reports/Q1-2026-Support-Cases.xlsx", drive="team-shared")
"Report uploaded to shared drive: /Reports/Q1-2026-Support-Cases.xlsx"
9
Email Executive Summary
Microsoft Graph
"Email the summary to my VP of Support"
send_email(to="vp-support@company.com", subject="Q1 2026 Support Case Report", body=executive_summary, attachment="Q1-2026-Support-Cases.xlsx")
"Email sent to VP of Support with executive summary and Excel attachment."
10
Notify Team
Teams / Slack
"All done! Q1 report with 847 cases across 5 products. Key alert: Mobile SDK crash spike in Feb. Report in shared drive and emailed to VP. 3 charts generated. Anything else?"
What Makes This Possible in ContextWeaver
| Capability | How It's Used |
|---|---|
| Salesforce OAuth via Vault | Client ID + secret stored in vault. Backend authenticates using @vault:cw-org-salesforce-client-secret — user never handles OAuth tokens |
| Multi-connector plugin | One plugin wires Salesforce + Excel/OneDrive + Email + Teams into a single conversational workflow |
| SOQL query tool | AI constructs Salesforce SOQL queries dynamically based on the user's natural language request |
| Group-scoped connectors | Sales team sees customer cases; Support sees support cases; Managers see all — same plugin, different data via connector extensions |
| Agentic reasoning | AI doesn't just dump data — it analyzes trends, identifies spikes, ranks accounts, and generates executive summaries |
| Excel generation tool | Creates multi-sheet workbooks with formatting, filters, freeze panes, and embedded charts |
| MCP Prompts | quarterly-report prompt template guides the multi-step data extraction and analysis flow |
| Voice + follow-up | User asks follow-up questions mid-flow ("which customers have critical cases?") and the agent adapts |
Can Anyone Else Do This Today?
| Platform | Verdict | Why Not |
|---|---|---|
| Salesforce Einstein | ⚠️ Partial | Can query cases and show dashboards inside Salesforce, but can't export to Excel with custom charts, email externally, or chain with OneDrive — locked to SF ecosystem |
| ChatGPT + Code Interpreter | ⚠️ Partial | Can analyze CSV data and make charts, but can't authenticate to Salesforce, no vault, no RBAC, user must manually export data first |
| Power Automate + Power BI | ⚠️ Partial | Can build SF-to-Excel flows, but pre-built and rigid — no conversational follow-ups, no voice, no agentic reasoning, no dynamic SOQL |
| Claude + MCP | ⚠️ Partial | MCP tools yes, but single-user — you'd hardcode SF credentials, no team scoping, no vault, no Excel generation built-in |
| Zapier / Make | ⚠️ Limited | Can move data between systems but no AI analysis, no trend detection, no conversational follow-ups, no dynamic queries |
 ContextWeaver |
✅ Full Flow | Vault-backed SF OAuth, dynamic SOQL, AI trend analysis, Excel with charts, OneDrive upload, email delivery, team notifications — all conversational with follow-ups |
The key differentiator: No other platform lets you conversationally authenticate to Salesforce, query cases with AI-generated SOQL, analyze trends, generate formatted Excel workbooks with charts, upload to SharePoint, and email the VP — all in one voice conversation with vault-backed credentials and team-scoped access control.
Connectors Used in This Scenario
☁️
Salesforce
SOQL queries, case data
🔒 OAuth client secret in vault
📊
Excel / OneDrive
Workbook + charts + upload
🔒 Graph API token in vault
📧
Email (Graph)
Send with attachments
🔒 Delegated permission in vault
💬
Teams / Slack
Notifications
🔒 Webhook URL in vault
All credentials stored in vault with team-scoped connector extensions — Sales sees customer cases, Support sees support cases, Managers see everything.
Team-Scoped Data Isolation
Same plugin, same tools, different data — powered by connector extensions:
| Team | Salesforce Filter | Sees |
|---|---|---|
| Sales Ops | RecordType = 'Customer Inquiry' | Customer inquiries, renewals, upsell cases |
| Support Engineering | RecordType = 'Technical Support' | Bugs, crashes, technical escalations |
| VP / Managers | (all records) | Everything — manager role override |
Each team's Salesforce connector extension has different SOQL filters — the same sf_query tool returns different data based on who's asking.