-
Notifications
You must be signed in to change notification settings - Fork 7
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
1 parent
25da05e
commit 65c93ab
Showing
6 changed files
with
293 additions
and
3 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,5 +1,5 @@ | ||
| --- | ||
| sidebar_position: 2 | ||
| sidebar_position: 1 | ||
| displayed_sidebar: mockingbird | ||
| --- | ||
|
|
||
|
|
||
2 changes: 1 addition & 1 deletion
2
apps/docusaurus/zeusfyi/docs/mockingbird/builder/_category_.json
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
8 changes: 8 additions & 0 deletions
8
apps/docusaurus/zeusfyi/docs/mockingbird/quickstart/_category_.json
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,8 @@ | ||
| { | ||
| "label": "Quickstart", | ||
| "position": 1, | ||
| "link": { | ||
| "type": "generated-index", | ||
| "description": "How to get started with a simple example" | ||
| } | ||
| } |
138 changes: 138 additions & 0 deletions
138
apps/docusaurus/zeusfyi/docs/mockingbird/quickstart/book_scoring.md
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,138 @@ | ||
| --- | ||
| sidebar_position: 1 | ||
| displayed_sidebar: mockingbird | ||
| --- | ||
|
|
||
| # Book Scoring | ||
|
|
||
| ## Overview | ||
|
|
||
| We'll be having the AI create a list of books, and then aggregating the titles into a JSON schema for scoring via an | ||
| Auto-Eval. | ||
| This pattern is also generally useful for synthetic data generation for ML training, and for creating a scoring system | ||
| for any type of data. | ||
|
|
||
| ## Step 1: Create an AI Analysis Task | ||
|
|
||
| ### Prompt: | ||
|
|
||
| Write a list of 10 book titles, with 5 of them being sci-fi, and 5 of them being any other genre. | ||
|
|
||
| ```json | ||
| [ | ||
| { | ||
| "title": "book_name", | ||
| "title": "book_name2" | ||
| } | ||
| ] | ||
| ``` | ||
|
|
||
|  | ||
|
|
||
| ## Step 2: Create a JSON schema | ||
|
|
||
| ### JSON Schema: | ||
|
|
||
| ```json | ||
| { | ||
| "book_scores": [ | ||
| { | ||
| "score": 9, | ||
| "title": "Dune" | ||
| }, | ||
| { | ||
| "score": 10, | ||
| "title": "Foundation" | ||
| }, | ||
| { | ||
| "score": 8, | ||
| "title": "Neuromancer" | ||
| } | ||
| ] | ||
| } | ||
| ``` | ||
|
|
||
| ### JSON Schema Builder | ||
|
|
||
|  | ||
|
|
||
| #### JSON Schema Fields | ||
|
|
||
| ##### Schema Name: book_scores | ||
|
|
||
| Field Name: score | ||
| Field Type: number | ||
|
|
||
| Field Description: Score each book from 1-10, with 10 being the | ||
| highest likelihood of being a science-fiction book. | ||
|
|
||
| Field Name: title | ||
| Field Type: string | ||
| Field Description: The title of the book. | ||
|
|
||
| After saving, you'll be able to see this schema in the schema list. | ||
|
|
||
|  | ||
|
|
||
| ## Step 4: Create an Aggregate Task | ||
|
|
||
| Since the JSON schema fields are already defined, we can now create an aggregation task to score the books. | ||
| We don't need to do any additional steps because the JSON schema field descriptions give the AI enough information to | ||
| score the books | ||
| based on the schema and field descriptions. | ||
|
|
||
|  | ||
|
|
||
| ## Step 5: Create an Eval Fn Task | ||
|
|
||
| Since the JSON schema fields are already defined, we can reuse this schema to create an Auto-Eval task to score the | ||
| books. | ||
| Set the criteria for the Auto-Eval to score passing if the book is rated > 2/5 for likelihood to be a sci-fi book, and | ||
| failing if it is not. | ||
|
|
||
|  | ||
|
|
||
| ## Step 6: Create the Workflow | ||
|
|
||
| ### Add Tasks to the Workflow | ||
|
|
||
| - Add a name and group for the workflow | ||
| - Add the analysis task to the workflow. | ||
| - Add the aggregation task to the workflow. | ||
| - Add the eval fn task to the workflow. | ||
|
|
||
| ### Connect Analysis -> Aggregation | ||
|
|
||
| - Connect the analysis task to the aggregation task. | ||
|
|
||
|  | ||
|
|
||
| ### Connect Eval Fn -> Aggregation | ||
|
|
||
| - Toggle to the aggregation option | ||
|
|
||
|  | ||
|
|
||
| - Connect the aggregation task to the Auto-Eval task. | ||
|  | ||
|
|
||
| ### Review and Save | ||
|
|
||
| - Review the workflow to ensure all tasks are connected properly. | ||
|  | ||
|
|
||
| Save the workflow. | ||
|
|
||
| ### Run the Workflow | ||
|
|
||
|  | ||
|
|
||
|  | ||
|
|
||
| ### Review the Results | ||
|
|
||
|  | ||
|
|
||
|  | ||
|
|
||
|
|
144 changes: 144 additions & 0 deletions
144
apps/docusaurus/zeusfyi/docs/mockingbird/quickstart/tweetflow.md
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,144 @@ | ||
| --- | ||
| sidebar_position: 1 | ||
| displayed_sidebar: mockingbird | ||
| --- | ||
|
|
||
| # Tweetflow | ||
|
|
||
| ## Overview | ||
|
|
||
| This guide is crafted to walk you through the process of developing a system that streamlines your social media | ||
| interactions, specifically tailored for Twitter using Zeusfyi’s Mockingbird AI system. | ||
|
|
||
| Our goal is to enable you to efficiently analyze and respond to your latest saved bookmarks on Twitter. The workflow | ||
| we’re constructing will leverage cutting-edge AI to curate context-aware tweet replies automatically. Here’s an overview | ||
| of what this guide will cover: | ||
|
|
||
| - Automated Analysis: The workflow begins by analyzing your most recent Twitter bookmark, extracting key themes and | ||
| sentiments to inform the response generation. | ||
| - Reply Generation: Using sophisticated AI, the system will then compose a reply that is relevant and engaging, ensuring | ||
| the conversation flows naturally. | ||
| - Auto-Evaluation: Before any tweet reaches you, it undergoes an automated evaluation process. This step assesses the | ||
| reply against pre-established quality criteria and metrics to ensure appropriateness and alignment with your digital | ||
| voice. | ||
| - Approval Mechanism: Once the tweet passes the auto-evaluation, it is forwarded to you for final review. This guide | ||
| will explain how to approve or reject the suggested content, giving you complete control over what gets posted. | ||
| - Execution: Upon your approval, the tweet will be automatically posted, maintaining timely and relevant engagement with | ||
| your audience. | ||
|
|
||
| - By following this tutorial, you will set up a workflow that not only speeds up your response time on Twitter but also | ||
| ensures that each interaction is thoughtful and brand-consistent. Let’s embark on this journey to enhance your Twitter | ||
| presence with automation and precision. | ||
|
|
||
| ## Step 1: Automated Twitter Auth & Routing Table Setup | ||
|
|
||
| Connect Twitter: Start by establishing authorization and a routing table for Twitter API calls. | ||
| Use the “Automated Twitter Auth & Routing Table Setup” to create a routing group named twitter-{YOUR_TWITTER_HANDLE}. | ||
| This will also generate a bearer token and save it in the platform's secret manager. | ||
| This section is in the indexer tab on the main AI page. It will redirect to Twitter for you to authorize your account | ||
| for usage. | ||
|
|
||
|  | ||
|
|
||
| ## Step 2: Inputs/Outputs Configuration | ||
|
|
||
| API via Load Balancer: Configure an API endpoint that can be used by a load balancer to aggregate data from different | ||
| routes. | ||
| This is for setting up general retrieval processes that can quickly start. In the below example. You’ll need to setup | ||
| the below two stages. | ||
|
|
||
| ### Input Retrieval: Getting Your Latest Tweet Bookmark | ||
|
|
||
| - Route: `users/:id/bookmarks?max_results=1` | ||
|
|
||
|  | ||
|
|
||
| ### Reply via API: Sending an API request to Twitter to post a Tweet. | ||
|
|
||
|  | ||
|
|
||
| ## Step 3: Defining a JSON Schema | ||
|
|
||
| Define the Twitter post request schema | ||
| Add Fields: Add field names, types, and descriptions necessary for the Twitter API (e.g., text, in_reply_to_tweet_id). | ||
|
|
||
| Note: Actual API request needed for creating tweet replies | ||
|
|
||
| ```json | ||
| { | ||
| "text": "tweet body text", | ||
| "reply": { | ||
| "in_reply_to_tweet_id": "tweet_id_value" | ||
| } | ||
| } | ||
| ``` | ||
|
|
||
| Simplified schema tweet API request we’re building | ||
|
|
||
| ```json | ||
| { | ||
| "text": "tweet body text", | ||
| "in_reply_to_tweet_id": "in_reply_to_tweet_id" | ||
| } | ||
| ``` | ||
|
|
||
| We can skip the nested object field for this specific request since we provide an embedded transformer which | ||
| converts to the above format when API requests are sent to the twitter API from a human-in-the-loop approved tweet. | ||
|
|
||
|  | ||
|
|
||
| ## Step 4: Creating the AI Task for Tweet Response | ||
|
|
||
| Add an analysis task to generate a tweet response based on the latest bookmarked tweet. | ||
|
|
||
|  | ||
|
|
||
| ## Step 5: Creating the Trigger Tweet Response API Call | ||
|
|
||
| You attach your previous API Input/Output stage for it to route your API request after it’s been approved. This also | ||
| only triggers an action for approval if all your eval info stages pass. | ||
|
|
||
|  | ||
|
|
||
|  | ||
|
|
||
| ## Step 6: Creating the Eval Task for Tweet Response | ||
|
|
||
|  | ||
|
|
||
| You’ll attach the JSON schema we created earlier and then set the eval criteria on the fields. In this example it | ||
| verifies that the text body meets the 280 character limit for twitter and that the tweet _id being replied to is not | ||
| empty. You then attach the trigger to execute on Eval completion. | ||
|
|
||
| ## Step 7: Creating the Workflow | ||
|
|
||
| - Attach bookmark retrieval | ||
| - Attach tweet response analysis task | ||
| - Attach eval to your tweet response analysis task | ||
| - Save workflow | ||
|
|
||
|  | ||
|
|
||
|  | ||
|
|
||
| ## Step 8: Run Execute Workflow | ||
|
|
||
| Since we just want to execute by cycles. Just select 1 cycle of runtime and then press start. | ||
| Temporal Workflow Integration: Use the durable execution feature of Temporal workflows to chain tasks together, running | ||
| them on a schedule or manually. | ||
|
|
||
|  | ||
|
|
||
| ## Run Inspection | ||
|
|
||
| You can view the results of your eval stages and tasks in the Run section. If your eval passed, you’ll be able to see | ||
| the trigger action in the Actions tab. | ||
|
|
||
|  | ||
|
|
||
| ## Step 9: Inspect and Approve AI Driven API Call | ||
|
|
||
| Now you can select Approve or Reject for the AI generated Tweet. If you like it, press approve and it will execute the | ||
| POST tweet request using the JSON body output from your Auto-Eval stage. | ||
|
|
||
|  |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters