Twelve reference patterns, from foundational workflows to multi-layer Azure and AI architectures. Each pairs a people view with a technical design, drawn in the format that explains it best: sequences, state models, journeys, timelines, Gantt plans and layered architectures.
Capability Map / ADVANCED PATTERN
Connected Expertise. One Delivery Model.
Timeline. How an engagement moves from discovery to ongoing improvement, with a clear finish point at each stage.
Layered architecture. How business analysis and project management connect Power Platform, AI services and Azure under shared governance.
Business analysis turns goals into requirements; project management plans and controls the delivery around them.
Power Platform, AI services and Azure are designed as connected layers rather than separate tools.
Shared identity, environment policies and runbooks keep the whole solution governable after launch, and measured outcomes feed the next round of requirements.
Design Considerations
A capability map, not a promise that every engagement needs every platform. Licensing and integrations depend on scope.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
AI & Automation / ADVANCED PATTERN
Extract the Detail. Keep People in Control.
Sequence diagram. Who hands what to whom, in order, when an invoice arrives—and what changes when the numbers do not match.
Layered architecture. Five layers from intake to audit, with confidence thresholds and tolerance rules deciding where people step in.
An authenticated integration retrieves approved invoice and timesheet documents for processing.
Document Intelligence extracts fields; Python matching logic compares contractor identities, hours and rates with the approved records.
Exceptions go to finance for review. Only approved items continue into the agreed downstream workflow.
Design Considerations
Based on the recruitment accounts-payable project using Document Intelligence, Python and MSAL. This is a simplified pattern; extraction is not treated as a payment approval or a guarantee of accuracy.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
AI & Automation / ADVANCED PATTERN
Find the Answer. Respect the Access Boundary.
User journey with satisfaction scores. How the experience feels at each step, scored from 1 to 5, including the moment an answer is not available.
Dual-pipeline architecture. A scheduled ingestion pipeline and a per-question query pipeline, joined by a permission-aware index and an evaluation loop.
Approved sources are chunked, embedded and indexed on a schedule, with access-control fields stored alongside each passage.
Each question is searched with the user’s own group claims as a security filter, then re-ranked for relevance.
The model answers only from retrieved passages, with citations. When evidence is thin, it says so and refers the question to an owner.
Design Considerations
An illustrative retrieval-augmented generation design informed by the case-file search project. Index design, refresh frequency and evaluation criteria are agreed per use case.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
AI & Automation / INTERMEDIATE PATTERN
Self-Service When It Helps. A Human When It Matters.
Sequence diagram. A real conversation pattern: resolved by guidance, or handed to a technician with the context already captured.
Component architecture. Channels, the agent runtime, the service desk and the improvement loop behind the reported 30% ticket reduction.
An employee describes the issue. The assistant retrieves relevant approved support material within the intended access boundaries.
The employee tries the guidance and confirms whether it helped. Missing or unsuitable answers are routed to a human rather than presented as certainty.
The handoff carries the issue and steps already attempted. Employee feedback and reporting inform a reviewed knowledge-improvement process.
Design Considerations
A reference flow based on the supplied employee support-agent project, which reported 30% fewer IT tickets over three months. Feedback does not automatically retrain a model; improvements require review and the agreed implementation process.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Power Platform / FOUNDATIONAL PATTERN
One Request. A Visible Decision.
User journey with satisfaction scores. The same request scored before and after the app, from the requester’s and manager’s point of view.
State diagram. Every state a request can be in, and the event that moves it on—including clarification and escalation.
A structured request captures the business reason and required details before routing.
A workflow applies the agreed approval rules and records decisions, with clarification and escalation paths.
Only approved requests move to fulfillment. Ownership, status and completion remain visible to the requester and delivery team.
Design Considerations
An illustrative reference pattern linked to the supplied hardware approval project. Platform choices, approvers, escalation periods, licensing and integrations require discovery. A workflow status does not itself authorize spending or procurement.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Power Platform / INTERMEDIATE PATTERN
Capture Once. Sync When Connected.
Process flow. The field worker’s loop: capture offline, keep working, sync when coverage returns.
Sequence diagram. What passes between the mobile app, the on-device store, Dataverse and Power Automate, including conflicts.
How It Works
Users sign in and prepare the app and permitted records while connected, before heading into the field.
The app stores supported changes locally while offline and displays their sync status.
When connectivity returns, changes synchronize and conflicts or errors are surfaced for resolution. Reporting uses synchronized records.
Design Considerations
An illustrative Power Apps / Dataverse option based on the supplied offline-app project. Offline support, data security and conflict handling must be designed and tested for the selected devices and app type.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Azure / ADVANCED PATTERN
A Landing Zone With Clear Boundaries.
Decision tree. How each workload is assessed: retire, rehost, refactor or keep on-premises under Azure Arc.
A hub network holds shared connectivity: the VPN gateway, firewall and Bastion for administrative access.
Workloads sit in a spoke network with separate web, application, integration and private-endpoint subnets, so each tier is reachable only from where it should be.
Key Vault, managed identities, monitoring, backup and budgets are shared platform services, governed by Azure Policy and role-based access.
Design Considerations
An illustrative reference pattern. The right topology, security services and SKUs depend on the workload, compliance needs and budget.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Azure / INTERMEDIATE PATTERN
A Consistent Desktop. A Planned Recovery Path.
Sequence diagram. A desktop incident from first report to business acceptance, with the two recovery paths that apply.
Component architecture. Identity, the AVD control plane, session hosts, the golden-image lifecycle and operations in one view.
Users connect to Azure Virtual Desktop through identity controls and the managed service connection path.
Session hosts run within the Azure network, with profile storage supporting the designed user experience.
A maintained golden image supports repeatable host deployment. A separate recovery plan defines owners, escalation, dependencies and scenario exercises.
Design Considerations
An illustrative Azure Virtual Desktop design informed by the desktop and disaster-recovery projects. A golden image alone is not disaster recovery; backups, profile protection, application data, recovery targets and restore tests require their own design.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Azure / FOUNDATIONAL PATTERN
On-Premises Power. Managed From Azure.
Process flow. Five steps from a compute need to a server governed alongside cloud resources.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Business Analysis / INTERMEDIATE PATTERN
From a Business Need to a Tested Requirement.
Mind map. What a discovery workshop captures before anyone discusses tools.
Traceability diagram. Each requirement traces from the business objective to user stories, acceptance tests and sign-off.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Project Management / INTERMEDIATE PATTERN
Phased Rollout. Controlled Change.
State diagram. How a change request moves from raised to delivered, deferred or rejected—without derailing the plan.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.
Migration / INTERMEDIATE PATTERN
Move the Content. Preserve the Way People Work.
Quadrant chart. Sorting content by use and value decides what moves first, what is archived and what is removed.
Phased process. Five phases with a pilot sign-off loop and wave-based migration.
Discover sites, content owners, permissions and dependencies. Identify outdated content, customizations and migration blockers before choosing the migration approach.
Design the destination information structure and access model. Run a representative pilot and validate content, metadata and permissions with business owners.
Migrate in agreed waves, reconcile changes and validate before cutover. Guide users to the new locations and hand over ownership and support documentation.
Design Considerations
A reference migration journey based on the supplied SharePoint migration project. Tool choice, supported metadata, custom solutions, retention requirements, delta migration and downtime must be assessed. Source decommissioning follows agreed validation and retention decisions.
Illustrative reference patterns, not exact client deployment maps. Use View Full Screen to fit the complete diagram, or scroll within this panel to inspect detail.