D365 AI Orchestration: Finance, SCM & Customer Experience
Orchestrate AI features across D365 Finance, Supply Chain Management, Sales, and Service. Extend F&O agent chat with custom knowledge sources and add help guidance for D365 apps.
AI is already built into D365 — your job is to orchestrate it
Think of D365 AI features as instruments in an orchestra. Each instrument (feature) can play solo, but the magic happens when you orchestrate them together.
D365 Finance has AI for cash flow predictions, budget proposals, and collections prioritisation. D365 SCM has AI for demand planning, supplier communications, and inventory optimisation. D365 Sales and Service have AI for summarising records, drafting emails, and routing cases.
Your job as a solution architect isn’t to build these features — they’re already there. Your job is to orchestrate them: turn on the right ones, configure them for your client’s data, connect them to custom knowledge sources, and make sure they work together.
AI features in D365 Finance
| Feature | What It Does | Business Value |
|---|---|---|
| Collections coordinator summary | AI summarises a customer’s payment history, overdue invoices, and contact history for the collections agent | Faster collections calls — agent has full context without digging through records |
| Workflow history summary | AI summarises the approval history of a document (who approved, when, what comments were made) | Quick decision-making on stalled workflows — see the full story at a glance |
| Budget proposal assistance | AI generates budget proposals based on historical spending patterns and forecasts | Faster budget cycles — draft proposals in minutes instead of days |
| Financial insights | AI analyses financial data to surface trends, anomalies, and recommendations | Proactive financial management — catch issues before they become problems |
| Cash flow forecasting | AI predicts future cash positions based on receivables, payables, and historical patterns | Better liquidity management — plan cash needs weeks in advance |
| Customer payment predictions | AI predicts when customers will pay invoices based on their payment history | Prioritise collections — focus on customers likely to pay late |
AI features in D365 Supply Chain Management
| Feature | What It Does | Business Value |
|---|---|---|
| Demand planning with Copilot | AI analyses historical demand, seasonality, and market signals to forecast future demand | Reduce stockouts and overstock — right inventory at the right time |
| Supplier communications agent | Autonomous agent that communicates with suppliers about PO confirmations, delivery updates, and discrepancies | Reduce manual email work — agent handles routine supplier correspondence |
| Workload insights | AI analyses production workload across resources and suggests rebalancing | Optimise factory utilisation — prevent bottlenecks before they happen |
| Inventory queries with Copilot | Natural language queries against inventory data (“What’s the stock level for part X in warehouse 3?”) | Faster inventory answers — no need to navigate complex reports |
| Confirmed purchase orders summary | AI summarises incoming PO confirmations and highlights changes from original orders | Quick exception handling — focus on what changed, not what’s routine |
AI features in D365 Sales and Service
| Feature | What It Does | Business Value |
|---|---|---|
| Record summarisation | AI summarises account, opportunity, case, and contact records in natural language | Instant context — understand a record’s full story without reading every field |
| Email assistance | AI drafts contextual email replies based on the record and conversation history | Faster responses — draft quality emails in seconds |
| Case routing with AI | Intelligent routing that analyses case content to assign to the best-qualified agent | Better first-contact resolution — cases reach the right expert immediately |
| Contact Centre agents | AI-powered agents that handle customer interactions across voice, chat, and digital channels | Scale customer service — handle more interactions without more headcount |
| Copilot for opportunity management | AI provides insights, next-best actions, and risk alerts for sales opportunities | Proactive deal management — close more deals with AI-guided selling |
| Knowledge article suggestions | AI recommends relevant knowledge articles to service agents during case resolution | Faster resolution — agents find answers without searching the knowledge base |
Extending F&O agent chat with additional knowledge
The built-in agent chat in D365 Finance and Operations apps provides generative help and guidance that is grounded in Microsoft product documentation by default. You can extend it to include your organisation’s custom knowledge, but this is done through Copilot Studio — not through F&O admin settings directly.
How to add custom knowledge:
- Open the Copilot for Finance and Operations agent in Copilot Studio — this is the agent that powers the in-app chat experience
- Add custom topics — create topics that connect to your organisation’s knowledge sources (SharePoint document libraries, internal wikis, company policies)
- For structured F&O data — use virtual entities or Dataverse tables as knowledge sources so the agent can query live business data
- For unstructured content (procedures, policies) — add SharePoint or other document sources as knowledge within the Copilot Studio topic
- Test and validate — verify the agent returns accurate answers from your custom sources before rolling out
Scenario: Kai extends F&O chat for Apex's internal procedures
Kai Mercer extends the D365 SCM agent chat at Apex Industries. The default chat answers product documentation questions well — “how to create a purchase order” — but can’t answer organisation-specific questions like “what’s Apex’s approval threshold for capital expenditure?”
Extension design:
Step 1 — Identify knowledge gaps: Kai surveys 20 SCM users. Top questions the default chat can’t answer:
- “What’s our vendor approval process?”
- “What are the safety stock minimums for Class A parts?”
- “How do we handle returns from international suppliers?”
Step 2 — Prepare knowledge sources:
- Priya Sharma (data engineer) organises Apex’s internal procedures in a SharePoint document library: procurement policies, safety stock rules, returns procedures
- Documents are cleaned, deduplicated, and tagged by department
Step 3 — Extend via Copilot Studio:
- Kai opens the Copilot for Finance and Operations agent in Copilot Studio
- Creates a custom topic that retrieves answers from the SharePoint document library
- For live data queries (e.g. “what’s the current stock level for part X?”), configures a topic that queries Dataverse virtual entities mapped to F&O tables
Step 4 — Test: Tomasz Kowalski (D365 consultant) tests 30 questions against the extended chat. 27 return accurate answers from the custom sources. 3 need document improvements.
Result: SCM users get instant answers about Apex-specific procedures directly in D365 SCM — no need to leave the app or search SharePoint manually.
Adding knowledge to in-app help and guidance
In-app help and guidance in D365 Finance and SCM provides contextual help panels within the application. You can extend these with custom knowledge sources:
Process:
- Author content — create task guides, help articles, and procedure documentation in your custom knowledge source
- Structure for context — tag content by D365 module, form, or process so the help system can match it to the user’s current screen
- Connect the source — configure the help system to include your custom documentation alongside standard Microsoft help
- Set priority — determine whether custom content appears before or alongside Microsoft’s default help
- Maintain — update custom help content when processes change. Stale help is worse than no help
| Feature | What It Covers | Who Uses It | Governance |
|---|---|---|---|
| D365 prebuilt AI | Standard AI features shipped with D365 — summarisation, predictions, routing, insights | End users in their daily workflow — the AI is embedded in the app experience | Enabled/disabled by admin. Configure data sources and scope. Limited customisation of the AI logic itself |
| Custom AI (Foundry + extensions) | Domain-specific AI beyond prebuilt — custom models, extended agents, additional knowledge sources | Same end users, but with enhanced AI capabilities tailored to the organisation's needs | Full control — you design, deploy, monitor, and maintain. Requires ML/AI expertise and ongoing investment |
Exam tip: memorise the feature-to-app mapping
The exam frequently asks “Which D365 app provides this AI feature?” Here’s the quick reference:
D365 Finance: Collections coordinator summary, workflow history summary, budget proposal assistance, financial insights, cash flow forecasting, customer payment predictions
D365 SCM: Demand planning with Copilot, supplier communications agent, workload insights, inventory queries, confirmed PO summary
D365 Sales: Record summarisation, email assistance, opportunity management insights
D365 Customer Service: Case routing with AI, Contact Centre agents, knowledge article suggestions, record summarisation, email assistance
Both Sales and Service: Record summarisation and email assistance appear in both — but case routing and Contact Centre agents are Service only.
Don’t confuse “supplier communications agent” (SCM — talks to suppliers) with “Contact Centre agents” (Service — talks to customers).
Flashcards
Knowledge check
Kai's client Apex Industries wants to reduce manual work in their procurement team. Currently, a procurement specialist spends 3 hours daily emailing suppliers to confirm purchase orders, check delivery dates, and resolve discrepancies. Which D365 SCM AI feature should Kai configure?
Adrienne's customer service team at Vanguard uses D365 Customer Service. Service agents report that finding the right knowledge article during a customer call takes too long — they search manually and often miss relevant articles. Which D365 AI feature addresses this?
A D365 Finance user asks the built-in agent chat: 'What is our company's capital expenditure approval process?' The chat returns a generic Microsoft Learn article about D365 approval workflows instead of the company's specific policy. What should the solution architect recommend?
🎬 Video coming soon
🏆 Domain 2 complete! You’ve covered the full breadth of designing AI-powered business solutions — from Copilot Studio conversation design to D365 AI orchestration. The exam tests your ability to match the RIGHT tool to the RIGHT scenario, choose the simplest extensibility path that works, and orchestrate AI features across the Microsoft ecosystem.
Next up: Agent Monitoring & Performance — moving into Domain 3: Implement and Manage AI Solutions, starting with how to monitor, measure, and optimise the AI solutions you’ve designed.