Copilot in D365 Customer Experience & Service
Dynamics 365 Sales, Customer Service, and Contact Center all have Copilot capabilities that can be customised and extended. Learn how to design business terms, customise Copilot behaviour, configure connectors, and integrate agents with Contact Center channels.
AI inside the apps your teams already use
Imagine Copilot is a smart assistant sitting inside every Dynamics 365 app. In Sales, it summarises deals and drafts emails. In Customer Service, it suggests responses and summarises cases. In Contact Center, it routes calls to the right agent — human or AI.
But here’s the thing: out-of-the-box Copilot doesn’t know YOUR business. It doesn’t know that when your team says “high-value client” they mean someone with over $1M in annual contracts. It doesn’t know that “escalation” means routing to the VP, not just the team lead.
As an architect, you customise Copilot so it speaks your business language, connects to your data sources, and integrates seamlessly with your contact centre channels.
Designing business terms for Copilot
Business terms in D365 Copilot let you teach Copilot your organisation’s vocabulary. When a user asks Copilot about “premium accounts,” the business terms definition tells Copilot exactly what that means in your context. This capability is primarily available in D365 Sales today, where it helps Copilot interpret sales-specific queries using your custom definitions.
| Element | What You Define | Example |
|---|---|---|
| Term name | The business concept | ”High-value customer” |
| Definition | What it means in your organisation | ”A customer with annual contract value over $500K and NPS score above 8” |
| Related entities | Which D365 tables and fields it maps to | Account table, Annual Revenue field > $500K AND NPS field > 8 |
| Synonyms | Other ways users might refer to the concept | ”Premium client,” “strategic account,” “key account” |
Scenario: Adrienne defines business terms for Vanguard's sales team
Adrienne configures business terms for Vanguard Financial Group’s D365 Sales Copilot:
- “Strategic account” — A client with assets under management over $10M and at least 3 active product lines (mapped to Account AUM field > $10M AND Active Products count >= 3)
- “At-risk deal” — An opportunity where the close date has been pushed back more than twice and competitor engagement is flagged
- “Warm intro” — A referral from an existing client relationship, not a cold outreach
Why it matters: Without these terms, Copilot interprets “show me strategic accounts” using generic logic. With business terms, it returns exactly the accounts Adrienne’s team needs to see.
Note: For D365 Customer Service, similar customisation goals are achieved through knowledge base configuration, case routing rules, and response suggestion policies rather than the business terms feature.
Customising Copilot in D365 CX and Service
Beyond business terms, architects can customise Copilot’s behaviour:
| What You Customise | How | Impact |
|---|---|---|
| Response suggestions | Configure knowledge sources, adjust response policies | Copilot suggests more relevant, on-brand responses to service agents |
| Case summarisation | Define which fields to include, summary format, key highlights | Agents get concise, relevant case summaries instead of generic ones |
| Email drafting | Set tone guidelines, template preferences, auto-include fields | Copilot drafts emails that match your brand voice and include required information |
| Record catch-up | Configure which activities, notes, and timeline entries to surface | Users see the most relevant recent activity when viewing a record |
| Agent scripts | Create guided workflows that Copilot follows during interactions | Consistent customer experience across all agents, human or AI |
Connectors for Copilot in D365 Sales
D365 Sales Copilot can be extended with connectors that bring in data from external systems:
- CRM enrichment connectors — Pull customer data from external CRMs (e.g., a subsidiary’s Salesforce instance)
- Market intelligence connectors — Feed competitor and industry data into deal analysis
- Communication connectors — Integrate email, Teams, and phone data for relationship insights
- ERP connectors — Connect to finance/inventory data for accurate quoting
Design considerations:
- Authentication: OAuth 2.0, API keys, or service accounts
- Data freshness: Real-time API calls vs scheduled sync
- Performance: Connector latency affects Copilot response time
- Security: Data classification of external data flowing into D365
Deep dive: Sales and Service in Microsoft 365 Copilot
Two specialised Copilot experiences extend D365 capabilities into M365 apps:
Sales in Microsoft 365 Copilot (formerly Copilot for Sales) — Embeds D365 Sales intelligence into Outlook and Teams. Auto-captures meeting notes, updates CRM records, generates email drafts with deal context, and provides real-time conversation tips during Teams calls.
Service in Microsoft 365 Copilot (formerly Copilot for Service) — Embeds D365 Customer Service intelligence into Outlook and Teams. Service agents can access case data, knowledge articles, and AI-generated responses without leaving their current app.
Architect’s decision: Configure these when your sales/service teams spend more time in Outlook/Teams than in D365. The data flows bidirectionally — Copilot reads from D365 and writes updates back.
Agents for D365 Contact Center
D365 Contact Center supports AI agents across multiple channels. The architect designs which channels agents serve and how they integrate.
| Channel | Agent Integration | Design Consideration |
|---|---|---|
| Voice | IVR replacement, natural language call routing, voice-enabled agents | Speech-to-text quality, voice agent persona, escalation to human |
| Live chat | Web chat bots, proactive chat triggers, co-browse support | Response latency, typing indicators, handoff experience |
| Social media | Facebook, Twitter/X, Instagram message handling | Tone adaptation, public vs private response routing |
| Auto-triage, suggested responses, auto-resolution for common queries | Classification accuracy, template management, SLA tracking | |
| Microsoft Teams | Internal help desk agents, employee self-service | Integration with M365 identity, access to internal knowledge |
Exam tip: channel-agent design principles
When the exam asks about designing agents for Contact Center channels:
- Start with the highest-volume channel — deploy agents where they’ll have the biggest impact first
- Design for escalation — every AI agent must have a clear path to a human agent
- Channel-appropriate tone — a voice agent sounds different from a chat agent
- Unified customer context — the agent should access the same customer record regardless of channel
- Omnichannel continuity — if a customer starts on chat and calls later, the agent should know the history
Flashcards
Knowledge check
A sales team at Vanguard Financial Group uses D365 Sales. Their account managers spend 70% of their time in Outlook and Teams, not in D365. They want Copilot to help with email drafting and meeting preparation using CRM data. What should Adrienne configure?
Jordan needs to deploy AI agents in CareFirst's contact centre across voice, chat, and email channels. What is the most important design principle Jordan should follow?
🎬 Video coming soon
Next up: Agent Types: Task, Autonomous & Prompt/Response — the three agent architectures you need to know, when to use each, and how they differ in capability and governance.