docs: Add session deliverables and comprehensive documentation updates

README.md

@@ -212,11 +212,19 @@ flowchart TB
     %% MODEL LAYER
     %% ═══════════════════════════════════════════════════════════════════════
     subgraph Models["Model Layer (VLMs)"]
-        direction LR
-        CLAUDE["Claude"]
-        GPT["GPT-4o"]
-        GEMINI["Gemini"]
-        QWEN["Qwen-VL"]
+        direction TB
+        subgraph APIModels["API Models"]
+            direction LR
+            CLAUDE["Claude"]
+            GPT["GPT-4o"]
+            GEMINI["Gemini"]
+        end
+        subgraph OpenSource["Open Source / Fine-tuned"]
+            direction LR
+            QWEN3["Qwen3-VL"]
+            UITARS["UI-TARS"]
+            OPENCUA["OpenCUA"]
+        end
     end
 
     %% ═══════════════════════════════════════════════════════════════════════

docs/architecture.md

@@ -79,26 +79,26 @@ flowchart TB
 
 ```mermaid
 flowchart LR
-    subgraph Record["1. Record"]
-        A[User Demo] --> B[Capture Session]
-        B --> C[Screenshots + Events]
+    subgraph Demonstrate["1. Demonstrate"]
+        A[Human Trajectory] --> B[Capture Session]
+        B --> C[Observations + Actions]
     end
 
     subgraph Store["2. Store"]
         C --> D[JSON/Parquet Files]
-        D --> E[Demo Library]
+        D --> E[Demonstration Library]
     end
 
-    subgraph Train["3. Train"]
-        E --> F[Data Loading]
-        F --> G[Model Training]
+    subgraph Learn["3. Learn"]
+        E --> F[Trajectory Abstraction]
+        F --> G[Policy Learning]
         G --> H[Checkpoint]
     end
 
-    subgraph Deploy["4. Deploy"]
-        H --> I[Agent Policy]
+    subgraph Execute["4. Execute"]
+        H --> I[Trained Policy]
         I --> J[Inference]
-        J --> K[Action Replay]
+        J --> K[Agent Deployment]
     end
 
     subgraph Evaluate["5. Evaluate"]
@@ -164,17 +164,17 @@ graph TD
 
 | Package | Responsibility | Key Exports |
 |---------|---------------|-------------|
-| **openadapt-capture** | GUI recording, event capture, storage | `CaptureSession`, `Recorder`, `Action` |
-| **openadapt-ml** | Model training, inference, adapters | `QwenVLAdapter`, `Trainer`, `AgentPolicy` |
+| **openadapt-capture** | Demonstration collection, observation-action capture, storage | `CaptureSession`, `Recorder`, `Action` |
+| **openadapt-ml** | Policy learning, training, inference | `QwenVLAdapter`, `Trainer`, `AgentPolicy` |
 | **openadapt-evals** | Benchmark evaluation, metrics | `ApiAgent`, `BenchmarkAdapter`, `evaluate_agent_on_benchmark` |
-| **openadapt-viewer** | HTML visualization, replay viewer | `PageBuilder`, `HTMLBuilder` |
+| **openadapt-viewer** | Trajectory visualization | `PageBuilder`, `HTMLBuilder` |
 
 ### Optional Packages
 
 | Package | Responsibility | Use Case |
 |---------|---------------|----------|
-| **openadapt-grounding** | UI element localization | Improved click accuracy with element detection |
-| **openadapt-retrieval** | Multimodal demo search | Find similar demonstrations for few-shot prompting |
+| **openadapt-grounding** | UI element grounding | Improved action accuracy with element detection |
+| **openadapt-retrieval** | Multimodal trajectory search | Find similar demonstrations for few-shot policy learning |
 | **openadapt-privacy** | PII/PHI scrubbing | Redact sensitive data before storage/training |
 
 ## Evaluation Loop
@@ -275,14 +275,14 @@ graph LR
 pip install openadapt
 
 # Individual packages
-pip install openadapt[capture]     # GUI capture/recording
-pip install openadapt[ml]          # ML training and inference
+pip install openadapt[capture]     # Demonstration collection
+pip install openadapt[ml]          # Policy learning and inference
 pip install openadapt[evals]       # Benchmark evaluation
-pip install openadapt[viewer]      # HTML visualization
+pip install openadapt[viewer]      # Trajectory visualization
 
 # Optional packages
-pip install openadapt[grounding]   # UI element localization
-pip install openadapt[retrieval]   # Demo search/retrieval
+pip install openadapt[grounding]   # UI element grounding
+pip install openadapt[retrieval]   # Trajectory retrieval
 pip install openadapt[privacy]     # PII/PHI scrubbing
 
 # Bundles

docs/cli.md

@@ -42,11 +42,11 @@ This verifies:
 
 ## Capture Commands
 
-Commands for recording user demonstrations.
+Commands for collecting human demonstrations.
 
 ### capture start
 
-Start a new recording session.
+Start a new demonstration collection session.
 
 ```bash
 openadapt capture start --name <name> [options]
@@ -64,25 +64,25 @@ openadapt capture start --name <name> [options]
 **Examples:**
 
 ```bash
-# Basic recording
+# Basic demonstration collection
 openadapt capture start --name login-task
 
-# Recording without screenshots
+# Demonstration collection without screenshots
 openadapt capture start --name audio-task --no-screenshots
 
-# Recording with slower screenshot interval
+# Demonstration collection with slower screenshot interval
 openadapt capture start --name slow-task --interval 1.0
 ```
 
 ### capture stop
 
-Stop the current recording.
+Stop the current demonstration collection.
 
 ```bash
 openadapt capture stop
 ```
 
-Alternatively, press `Ctrl+C` in the recording terminal.
+Alternatively, press `Ctrl+C` in the capture terminal.
 
 ### capture list
 
@@ -103,7 +103,7 @@ form-fill      89       5m 42s     2026-01-14
 
 ### capture view
 
-Open the viewer for a capture.
+Open the trajectory viewer for a demonstration.
 
 ```bash
 openadapt capture view <name> [options]
@@ -113,13 +113,13 @@ openadapt capture view <name> [options]
 
 | Argument | Required | Description |
 |----------|----------|-------------|
-| `<name>` | Yes | Name of the capture to view |
+| `<name>` | Yes | Name of the demonstration to view |
 | `--port` | No | Server port (default: 8080) |
 | `--no-browser` | No | Don't open browser automatically |
 
 ### capture delete
 
-Delete a capture.
+Delete a demonstration.
 
 ```bash
 openadapt capture delete <name>
@@ -129,11 +129,11 @@ openadapt capture delete <name>
 
 ## Train Commands
 
-Commands for training ML models.
+Commands for policy learning from demonstrations.
 
 ### train start
 
-Start training a model on a capture.
+Start policy learning from a demonstration.
 
 ```bash
 openadapt train start --capture <name> --model <model> [options]
@@ -143,7 +143,7 @@ openadapt train start --capture <name> --model <model> [options]
 
 | Argument | Required | Description |
 |----------|----------|-------------|
-| `--capture` | Yes | Name of the capture to train on |
+| `--capture` | Yes | Name of the demonstration to train on |
 | `--model` | Yes | Model architecture |
 | `--epochs` | No | Number of training epochs (default: 10) |
 | `--batch-size` | No | Batch size (default: 4) |
@@ -159,10 +159,10 @@ openadapt train start --capture <name> --model <model> [options]
 **Examples:**
 
 ```bash
-# Basic training
+# Basic policy learning
 openadapt train start --capture login-task --model qwen3vl-2b
 
-# Training with custom parameters
+# Policy learning with custom parameters
 openadapt train start \
     --capture login-task \
     --model qwen3vl-7b \
@@ -173,7 +173,7 @@ openadapt train start \
 
 ### train status
 
-Check training progress.
+Check policy learning progress.
 
 ```bash
 openadapt train status
@@ -191,7 +191,7 @@ ETA: 15 minutes
 
 ### train stop
 
-Stop the current training.
+Stop the current policy learning.
 
 ```bash
 openadapt train stop