ad-agent_architecture 0.0.2 → 0.0.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +7 -0
- data/README.md +4 -0
- data/docs/requirements.md +99 -0
- data/lib/ad/agent_architecture/version.rb +1 -1
- data/package-lock.json +2 -2
- data/package.json +1 -1
- metadata +2 -1
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 21906e309a921a97bcf97f4a9855d295f254a967c5c910783b58a709a6f8b9bf
|
|
4
|
+
data.tar.gz: d968abaef05bec72cb51646f4296e7d447e92746906e238d1ec4183f7168aa30
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: cb05cd58eaf70c090ef6a513f238825c6960bbbc59a314a0803400ba48203f15e3fced4f23c10f7157a67fd2dad08496fc74215198c0c1c25bb2ac8c15f0ca02
|
|
7
|
+
data.tar.gz: 6826716f20f4774f78f63d1dd1b8b41f4f03861561b46ad69ba65ad3ba1c39ec65329d8454402d915aa956029b180beb6926925fa3c7bbca10c853956e11fe67
|
data/CHANGELOG.md
CHANGED
|
@@ -1,3 +1,10 @@
|
|
|
1
|
+
## [0.0.2](https://github.com/appydave/ad-agent_architecture/compare/v0.0.1...v0.0.2) (2024-06-26)
|
|
2
|
+
|
|
3
|
+
|
|
4
|
+
### Bug Fixes
|
|
5
|
+
|
|
6
|
+
* test ci ([9514d51](https://github.com/appydave/ad-agent_architecture/commit/9514d5172edecffb2baa84e1911104487b331a6b))
|
|
7
|
+
|
|
1
8
|
## [Unreleased]
|
|
2
9
|
|
|
3
10
|
## [0.1.0] - 2024-06-26
|
data/README.md
CHANGED
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
### Analysis and Clarification of Workflow Documents
|
|
2
|
+
|
|
3
|
+
The provided documents outline structured workflows for creating YouTube scripts and Medium articles using AI tools and human input. Here’s a detailed analysis based on the initial documents and the additional information provided:
|
|
4
|
+
|
|
5
|
+
### Key Insights from Both Documents
|
|
6
|
+
|
|
7
|
+
1. **Human-AI Collaboration**:
|
|
8
|
+
- The YouTube script workflow is a complete list of steps performed primarily by AI (ChatGPT), with minimal human input, passed through various prompts to generate the output.
|
|
9
|
+
- The Medium article workflow is similar but not fully documented, indicating the same concept for generating content from a transcript.
|
|
10
|
+
- These workflows are common in agent automation frameworks, where AI performs most tasks, with humans providing quality assurance.
|
|
11
|
+
|
|
12
|
+
2. **Parameter Collection**:
|
|
13
|
+
- Parameters are collected and built over time, starting with a simple project title (simple_title) for YouTube and a transcript for Medium articles.
|
|
14
|
+
- Static input parameters like brand and target audience are crucial for defining the content's direction.
|
|
15
|
+
|
|
16
|
+
3. **Workflow Phases**:
|
|
17
|
+
- Both workflows consist of distinct phases: research and script/article writing.
|
|
18
|
+
- The research phase involves gathering information and generating potential content elements (titles, factsheets, topics).
|
|
19
|
+
- The writing phase focuses on drafting, revising, and finalizing the content.
|
|
20
|
+
|
|
21
|
+
4. **Input Parameters**:
|
|
22
|
+
- Parameters range from simple (single values) to complex (arrays of values).
|
|
23
|
+
- Initial parameters are relatively simple but grow in complexity as the workflow progresses.
|
|
24
|
+
|
|
25
|
+
5. **Prompts and Outputs**:
|
|
26
|
+
- Prompts are human-written text files with input parameter placeholders.
|
|
27
|
+
- Outputs can be simple (text) or complex (arrays of values such as engaging_titles[], keywords[], topics[]).
|
|
28
|
+
- Complex outputs generate new parameters for further steps.
|
|
29
|
+
|
|
30
|
+
6. **Human Decision Points**:
|
|
31
|
+
- Certain steps require human decisions, such as selecting the focus_video_type from multiple generated topics.
|
|
32
|
+
- Human input is crucial for quality assurance and final content selection.
|
|
33
|
+
|
|
34
|
+
7. **Iterative and Parallel Processes**:
|
|
35
|
+
- Workflows often involve iterative steps, refining outputs until satisfactory results are achieved.
|
|
36
|
+
- Parallel workflows can be initiated to explore multiple options simultaneously, using the same initial parameters but diverging paths.
|
|
37
|
+
|
|
38
|
+
### Detailed Analysis Document
|
|
39
|
+
|
|
40
|
+
#### Document 1: YouTube Script Creation Workflow
|
|
41
|
+
|
|
42
|
+
- **Parameters**:
|
|
43
|
+
- Initial: simple_title (e.g., "Fotor AI tool")
|
|
44
|
+
- Dynamic: Built over time (basic_factsheet, video_types[], focus_video_type, expanded_factsheet, topics[], keywords[], engaging_titles[], basic_script, basic_transcript, transcript_qa)
|
|
45
|
+
|
|
46
|
+
- **Phases**:
|
|
47
|
+
- **Research Phase**:
|
|
48
|
+
1. Generate potential titles using YouTube search.
|
|
49
|
+
2. Create a detailed factsheet using web search and AI tools.
|
|
50
|
+
3. Identify video types based on factsheet.
|
|
51
|
+
4. Expand factsheet with focus video type.
|
|
52
|
+
5. Generate engaging titles, keywords, and topics.
|
|
53
|
+
- **Script Writing Phase**:
|
|
54
|
+
1. Create basic script.
|
|
55
|
+
2. Clean and revise transcript.
|
|
56
|
+
3. Fact-check and finalize transcript.
|
|
57
|
+
|
|
58
|
+
- **Human Involvement**:
|
|
59
|
+
- Decision on focus_video_type.
|
|
60
|
+
- Final review and quality assurance.
|
|
61
|
+
|
|
62
|
+
#### Document 2: Medium Article Creation Workflow
|
|
63
|
+
|
|
64
|
+
- **Parameters**:
|
|
65
|
+
- Initial: transcript (YouTube transcript)
|
|
66
|
+
- Dynamic: article_recomendations, target_audience, outline, article_first_draft, outline_first_draft, introductions[], outline_first_draft_updated_with_intro
|
|
67
|
+
|
|
68
|
+
- **Phases**:
|
|
69
|
+
- **Research Phase**:
|
|
70
|
+
1. Generate article recommendations from transcript.
|
|
71
|
+
2. Draft the article based on target audience and recommendations.
|
|
72
|
+
3. Create preliminary outline from transcript.
|
|
73
|
+
- **Writing Phase**:
|
|
74
|
+
1. Write the first draft following the outline.
|
|
75
|
+
2. Generate and critique various introductions.
|
|
76
|
+
3. Revise and update outline with selected introduction.
|
|
77
|
+
|
|
78
|
+
- **Human Involvement**:
|
|
79
|
+
- Selection of introduction.
|
|
80
|
+
- Final review and quality assurance.
|
|
81
|
+
|
|
82
|
+
### Additional Insights
|
|
83
|
+
|
|
84
|
+
1. **Parameter Complexity**:
|
|
85
|
+
- Parameters evolve from simple inputs to complex structures, incorporating arrays of values that influence subsequent steps.
|
|
86
|
+
|
|
87
|
+
2. **Iterative Processes**:
|
|
88
|
+
- Both workflows benefit from iterative refinement, ensuring high-quality outputs before moving to the next phase.
|
|
89
|
+
|
|
90
|
+
3. **Parallel Workflows**:
|
|
91
|
+
- Exploring multiple options simultaneously (e.g., different video types) can provide richer content and better decision-making.
|
|
92
|
+
|
|
93
|
+
4. **Independent Subtasks**:
|
|
94
|
+
- Certain elements (like generating introductions for Medium articles) can operate independently but feed back into the main workflow, enhancing flexibility and adaptability.
|
|
95
|
+
|
|
96
|
+
5. **Cyclic Steps**:
|
|
97
|
+
- Iterative cycles help refine outputs continuously until optimal results are achieved, demonstrating a robust approach to content creation.
|
|
98
|
+
|
|
99
|
+
This structured approach to content creation leverages AI for efficiency while ensuring human oversight for quality, making it a highly effective workflow for producing engaging YouTube scripts and Medium articles.
|
data/package-lock.json
CHANGED
|
@@ -1,12 +1,12 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "ad-agent_architecture",
|
|
3
|
-
"version": "0.0.
|
|
3
|
+
"version": "0.0.3",
|
|
4
4
|
"lockfileVersion": 3,
|
|
5
5
|
"requires": true,
|
|
6
6
|
"packages": {
|
|
7
7
|
"": {
|
|
8
8
|
"name": "ad-agent_architecture",
|
|
9
|
-
"version": "0.0.
|
|
9
|
+
"version": "0.0.3",
|
|
10
10
|
"devDependencies": {
|
|
11
11
|
"@klueless-js/semantic-release-rubygem": "github:klueless-js/semantic-release-rubygem",
|
|
12
12
|
"@semantic-release/changelog": "^6.0.3",
|
data/package.json
CHANGED
metadata
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: ad-agent_architecture
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 0.0.
|
|
4
|
+
version: 0.0.3
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- David Cruwys
|
|
@@ -47,6 +47,7 @@ files:
|
|
|
47
47
|
- Rakefile
|
|
48
48
|
- bin/console
|
|
49
49
|
- bin/setup
|
|
50
|
+
- docs/requirements.md
|
|
50
51
|
- lib/ad/agent_architecture.rb
|
|
51
52
|
- lib/ad/agent_architecture/version.rb
|
|
52
53
|
- package-lock.json
|