Docs

docs/system/Metering.md

@@ -0,0 +1,15 @@
+# Metering
+
+## Purpose
+
+LLMs and Embedding models often are billed resources, so we want to ensure usage is limited and capped as desired.  
+
+## Interface
+
+Creates a global llm usage collection
+
+## Key Details
+
+- Ensure accurate and real-time metrics collection.
+- Provide actionable insights based on collected metrics.
+- Maintain data privacy and security.

docs/system/chat/ConversationalUI.md

@@ -0,0 +1,13 @@
+# Conversational UI
+
+## Purpose
+
+Provide a conversational chat UI with message history and non-text display which can be embedded or accessed on a full page. 
+
+## Interface
+
+- Access via FAB
+- Dedicated Page
+  - Include conversation history
+- Messages should support both markdown and cards 
+

docs/system/chat/LLMs.md

@@ -0,0 +1,15 @@
+# Large Language Models (LLMs)
+
+## Purpose
+
+Provide a reusable set of LLMs that can be used by the system easily for the right tasks. It should be easy to provide custom LLMs to the interface, and use local models in development.
+
+## Access
+
+LLMs should be accessible under custom for usage, subject to [Metering](../Metering.md) limits
+
+
+## Built in models
+
+- ollama
+- openai

docs/system/chat/MessageCollection.md

@@ -0,0 +1,36 @@
+# Chat Message Collection
+
+## Purpose
+
+Stores each message in a conversation, making them available to use as context and display to the user
+
+## Interface
+
+```mermaid
+erDiagram
+    USER{
+        string id
+        string name
+}
+MESSAGE {
+    string id
+    string content
+        string role
+        string conversationID
+        relation user
+        relationships references
+        timestamp sent 
+    }
+
+
+    MESSAGE ||--o{ USER : user
+    
+```
+
+## Key Details
+
+- Collection Will be added to the payload configuration during plugin initialization.
+- Messages Will be readable only by the user who created it
+- Messages Supports roles: 'assistant' and 'user' and 'tools
+- References field can store links to relevant documents used
+- Timestamp allows for chronological ordering of messages

docs/system/semantic-search/EmbeddingModel.md

@@ -0,0 +1,32 @@
+# Embedding Model
+
+## Purpose
+
+Convert chunks of text into comparable vectors for indexing and semantic serach.
+
+## Integration
+
+Integrated with the VectorStore to automatically embed queries/index documents as they are added. Provided in the form of a class
+
+```ts
+
+class CustomEmbeddingModel extends EmbeddingModel {
+
+    private model = new ModelClient();
+
+    generateEmbeddings(texts: string[]): number[] {
+        return this.model.embed(text);
+    }
+
+    numberOfDimensions() {
+        return 768;
+    }
+
+}
+
+```
+
+## Included Integrations
+
+- Ollama
+- OpenAI

docs/system/semantic-search/IndexableFields.md

@@ -0,0 +1,40 @@
+# Indexable Fields
+
+## Purpose
+
+Easily define what to index and how to query it using Payload Fields. This enables using existing fields or creating hidden virtual ones. Indexes are partitioned for performance and automatically integrate via collection hooks.
+
+## Interface
+
+```mermaid
+flowchart
+    
+    subgraph Collection
+    direction TB
+     create[Create] --> change[After Change Hook ]
+     update[Update] --> change
+     delete[Delete] --> change
+
+    end
+    
+    search[Search]
+
+    subgraph VectorStore
+    direction TB
+     embedding[Embedding Model]
+     persist[Persistance]
+     
+     embedding <--> persist
+    end
+    search --> embedding
+    change --> embedding
+
+
+```
+
+## Key Details
+
+- Configure any field by name `collection.field`
+- Backfills existing fields
+- Updates only when field changes
+

docs/system/semantic-search/RelatedDocumentsField.md

@@ -0,0 +1,9 @@
+# Related Documents Field
+
+## Purpose
+
+Provide a Custom Field type that automatically suggests and links related documents based on semantic similarity.
+
+## Interface
+
+Configurable based on top X closest, or within X distance

docs/system/semantic-search/SearchInterface.md

@@ -0,0 +1,6 @@
+# Search Interface
+
+## Purpose
+
+Provide a search UI Element that hooks into semantic search.
+

docs/system/semantic-search/VectorStore.md

@@ -0,0 +1,37 @@
+# Vector Store
+
+## Purpose
+
+Efficiently store, index, and query  vector representations of content for semantic search and similarity comparisons.
+
+
+## Identifier
+
+```ts
+
+type Identifier = {
+    documentID: string
+    collection: string
+    field: string
+}
+
+```
+
+## Class Integration
+
+```ts
+
+class MyVectorStore extends VectorDB{
+
+    upsert:(id: Identifier, value:string) => Promise<void>
+    delete:(id: Identifier) => Promise<void>
+    search:(filter: Partial<Identifier>, query: string) => Promise<Identifier &{distance: number}>
+
+}
+
+```
+
+## Included Implementation
+
+LanceDB
+

docs/use-cases/documentation-chatbot.md

@@ -0,0 +1,5 @@
+# Documentation chatbot
+
+Give your users a friendly chatbot that's integrated with you CMS managed documentation, always up to date.
+
+## Analyze questions users ask

docs/use-cases/draft-new-documents.md

@@ -0,0 +1,9 @@
+# Draft new documents
+
+## Prefill fields on create from prompt
+
+- Use previous documents as few shot training
+
+## Backfill new fields on existing documents
+
+## Compare to other docs and alert on thigns

docs/use-cases/improve-seo.md

@@ -0,0 +1,7 @@
+# IImprove SEO
+
+Find gaps and opportunities for additional content, or where to merge content.
+
+## Visualize Posts
+
+## Compare with AHREFs

docs/use-cases/improved-search.md

@@ -0,0 +1 @@
+# Improve Search

docs/use-cases/personalized-email-graphics.md

@@ -0,0 +1 @@
+# Personalized email graphics

docs/use-cases/personalized-recommendations.md

@@ -0,0 +1,5 @@
+# Personalized Recommendations
+
+## Based on whats in your cart
+
+## Based on questions asked

docs/use-cases/related-posts.md

@@ -0,0 +1,3 @@
+# Related Posts
+
+Use embeddings to recommend related posts, products, etc.