You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: learn-pr/wwl/design-alm-process-ai-powered-business-solutions/includes/3-design-alm-process-copilot-studio-agents-connectors-actions.md
+14-14Lines changed: 14 additions & 14 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -4,7 +4,7 @@ This unit guides solution architects through designing an Application Lifecycle
4
4
5
5
Architects learn how to structure solution components into managed lifecycles, establish promotion and approval workflows, manage data and environment boundaries, and integrate development tooling—enabling a predictable, repeatable process that aligns with enterprise governance.
6
6
7
-
## 1. ALM Foundations for Copilot Studio
7
+
## ALM Foundations for Copilot Studio
8
8
9
9
Copilot Studio solutions typically include:
10
10
@@ -30,7 +30,7 @@ A mature ALM process ensures:
30
30
31
31
* Sustainable lifecycle for updates and deprecations
32
32
33
-
## 2. Recommended Environment Strategy
33
+
## Recommended Environment Strategy
34
34
35
35
### Establish at least three core environments:
36
36
@@ -64,11 +64,11 @@ A mature ALM process ensures:
64
64
65
65
Iterate & Author Test, Approve, QA Support & Monitor
66
66
67
-
## 3. ALM for Agents
67
+
## ALM for Agents
68
68
69
69
Agents should move through lifecycle stages with predictable governance:
70
70
71
-
### 3.1 Development Stage
71
+
### Development Stage
72
72
73
73
* Draft agent scope, intents, and behaviors
74
74
@@ -78,7 +78,7 @@ Agents should move through lifecycle stages with predictable governance:
78
78
79
79
* Test agent workflows with edge-case prompts
80
80
81
-
### 3.2 Testing Stage
81
+
### Testing Stage
82
82
83
83
* Validate reasoning quality and output patterns
84
84
@@ -88,7 +88,7 @@ Agents should move through lifecycle stages with predictable governance:
88
88
89
89
* Run regression tests on all agent topics
90
90
91
-
### 3.3 Production Stage
91
+
### Production Stage
92
92
93
93
* Deploy via managed solutions
94
94
@@ -98,7 +98,7 @@ Agents should move through lifecycle stages with predictable governance:
98
98
99
99
* Document change history
100
100
101
-
## 4. ALM for Connectors
101
+
## ALM for Connectors
102
102
103
103
Connectors enable Copilot to interact with systems.
104
104
@@ -118,7 +118,7 @@ Connectors enable Copilot to interact with systems.
Copy file name to clipboardExpand all lines: learn-pr/wwl/design-alm-process-ai-powered-business-solutions/includes/5-design-alm-process-custom-ai-models.md
+24-24Lines changed: 24 additions & 24 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -4,7 +4,7 @@ This unit provides solution architects with a structured, enterprise-grade Appli
4
4
5
5
Custom AI models introduce unique ALM challenges such as data drift, model drift, regulatory alignment, and high-impact deployment risks. This unit provides architects with an actionable framework for governing model evolution from ideation through retirement.
6
6
7
-
## 1. ALM Foundations for Custom AI Models
7
+
## ALM Foundations for Custom AI Models
8
8
9
9
### A strong ALM process ensures:
10
10
@@ -16,7 +16,7 @@ Custom AI models introduce unique ALM challenges such as data drift, model drift
16
16
17
17
***Operational Readiness**: Runtime monitoring, governance logs, and rollback plans ensure business resilience.
18
18
19
-
## 2. Environment Strategy for Model Development
19
+
## Environment Strategy for Model Development
20
20
21
21
Establishing a multi-environment design prevents configuration drift and ensures safe promotions.
22
22
@@ -36,35 +36,35 @@ Establishing a multi-environment design prevents configuration drift and ensures
36
36
37
37
→ [Production]
38
38
39
-
## 3. ALM Lifecycle for Custom AI Models
39
+
## ALM Lifecycle for Custom AI Models
40
40
41
41
The lifecycle includes the following stages:
42
42
43
-
### 1. Plan & Design
43
+
### Plan & Design
44
44
45
45
* Define business use case, success criteria, constraints.
46
46
47
47
* Identify necessary data sources, governance rules, and privacy boundaries.
48
48
49
49
* Document intended model behavior, limitations, failure paths.
50
50
51
-
### 2. Data Preparation
51
+
### Data Preparation
52
52
53
-
*Build data contracts and schema standards.
53
+
* Build data contracts and schema standards.
54
54
55
-
*Establish curated datasets and "golden" evaluation datasets.
55
+
* Establish curated datasets and "golden" evaluation datasets.
56
56
57
-
*Use versioned data pipelines to prevent drift and ensure reproducibility.
57
+
* Use versioned data pipelines to prevent drift and ensure reproducibility.
58
58
59
-
### 3. Model Development
59
+
### Model Development
60
60
61
-
*Build, fine-tune, or incorporate "bring your own model" (BYOM) components.
61
+
* Build, fine-tune, or incorporate "bring your own model" (BYOM) components.
62
62
63
63
* Apply responsible AI requirements such as fairness, clarity, consistency, and safe response patterns.
64
64
65
65
* Maintain metadata: hyperparameters, training data versions, environment configuration snapshots.
66
66
67
-
### 4. Evaluation & Approval
67
+
### Evaluation & Approval
68
68
69
69
* Validate accuracy, relevance, reliability, and failure handling.
70
70
@@ -74,23 +74,23 @@ The lifecycle includes the following stages:
74
74
75
75
* Produce a **Model Card** containing metrics, constraints, and recommended usage boundaries.
76
76
77
-
### 5. Deployment
77
+
### Deployment
78
78
79
79
* Promote model via managed deployment artifacts.
80
80
81
81
* Apply model registries with version locking and rollback paths.
82
82
83
83
* Enforce access control, encryption, and runtime audit logging.
0 commit comments