diff --git a/Cargo.toml b/Cargo.toml
index 336d92b..9539a86 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -39,6 +39,7 @@ sha2 = "0.10"
 # Serialization
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
+base64 = "0.22"
 
 # Time management
 chrono = { version = "0.4", features = ["serde"] }
diff --git a/src/ai/mod.rs b/src/ai/mod.rs
index e4eaaad..163ec11 100644
--- a/src/ai/mod.rs
+++ b/src/ai/mod.rs
@@ -1,6 +1,10 @@
 /// AI/ML module with NPU support
 pub mod classifier;
 pub mod npu;
+pub mod models;
+pub mod vision;
 
 pub use classifier::NpuClassifier;
 pub use npu::NpuDevice;
+pub use models::{AvailableModels, ModelConfig, ModelDownloader};
+pub use vision::{ImageAnalyzer, ImageAnalysis};
diff --git a/src/ai/models.rs b/src/ai/models.rs
new file mode 100644
index 0000000..64ce9ff
--- /dev/null
+++ b/src/ai/models.rs
@@ -0,0 +1,149 @@
+/// AI Model management and downloading
+use std::path::PathBuf;
+use std::fs;
+use crate::error::{Result, AppError};
+
+/// Model configuration
+#[derive(Debug, Clone)]
+pub struct ModelConfig {
+    pub name: String,
+    pub url: String,
+    pub filename: String,
+    pub size_mb: u64,
+    pub description: String,
+}
+
+/// Available models
+pub struct AvailableModels;
+
+impl AvailableModels {
+    /// Mistral-7B-Instruct quantized (Q4) - Optimized for NPU
+    /// Size: ~4GB, good balance between quality and speed
+    pub fn mistral_7b_q4() -> ModelConfig {
+        ModelConfig {
+            name: "mistral-7b-instruct-q4".to_string(),
+            url: "https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF/resolve/main/mistral-7b-instruct-v0.1.Q4_K_M.gguf".to_string(),
+            filename: "mistral-7b-instruct-q4.gguf".to_string(),
+            size_mb: 4368,
+            description: "Mistral 7B Instruct Q4 - Text analysis and classification".to_string(),
+        }
+    }
+
+    /// CLIP for image understanding
+    pub fn clip_vit() -> ModelConfig {
+        ModelConfig {
+            name: "clip-vit-base".to_string(),
+            url: "https://huggingface.co/openai/clip-vit-base-patch32/resolve/main/onnx/model.onnx".to_string(),
+            filename: "clip-vit-base.onnx".to_string(),
+            size_mb: 350,
+            description: "CLIP ViT - Image and text embeddings".to_string(),
+        }
+    }
+
+    /// MiniLM for lightweight text embeddings
+    pub fn minilm() -> ModelConfig {
+        ModelConfig {
+            name: "minilm-l6".to_string(),
+            url: "https://huggingface.co/optimum/all-MiniLM-L6-v2/resolve/main/onnx/model.onnx".to_string(),
+            filename: "minilm-l6.onnx".to_string(),
+            size_mb: 90,
+            description: "MiniLM L6 - Fast text embeddings for classification".to_string(),
+        }
+    }
+}
+
+/// Model downloader
+pub struct ModelDownloader {
+    models_dir: PathBuf,
+}
+
+impl ModelDownloader {
+    pub fn new() -> Result<Self> {
+        let models_dir = PathBuf::from("models");
+        if !models_dir.exists() {
+            fs::create_dir_all(&models_dir)?;
+        }
+
+        Ok(Self { models_dir })
+    }
+
+    /// Check if a model is already downloaded
+    pub fn is_downloaded(&self, config: &ModelConfig) -> bool {
+        let model_path = self.models_dir.join(&config.filename);
+        model_path.exists()
+    }
+
+    /// Get the path to a model
+    pub fn get_model_path(&self, config: &ModelConfig) -> PathBuf {
+        self.models_dir.join(&config.filename)
+    }
+
+    /// Download a model (placeholder - requires actual HTTP client)
+    pub fn download(&self, config: &ModelConfig) -> Result<PathBuf> {
+        let model_path = self.models_dir.join(&config.filename);
+
+        if self.is_downloaded(config) {
+            log::info!("Model already downloaded: {}", config.name);
+            return Ok(model_path);
+        }
+
+        log::info!("Downloading model: {} ({} MB)", config.name, config.size_mb);
+        log::info!("URL: {}", config.url);
+        log::info!("Please download manually to: {}", model_path.display());
+
+        Err(AppError::Analysis(format!(
+            "Manual download required:\n\
+             1. Download from: {}\n\
+             2. Save to: {}\n\
+             3. Run again",
+            config.url,
+            model_path.display()
+        )))
+    }
+
+    /// List downloaded models
+    pub fn list_downloaded(&self) -> Result<Vec<String>> {
+        let mut models = Vec::new();
+
+        if let Ok(entries) = fs::read_dir(&self.models_dir) {
+            for entry in entries.flatten() {
+                if let Ok(file_type) = entry.file_type() {
+                    if file_type.is_file() {
+                        if let Some(filename) = entry.file_name().to_str() {
+                            models.push(filename.to_string());
+                        }
+                    }
+                }
+            }
+        }
+
+        Ok(models)
+    }
+}
+
+impl Default for ModelDownloader {
+    fn default() -> Self {
+        Self::new().expect("Failed to create ModelDownloader")
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_model_configs() {
+        let mistral = AvailableModels::mistral_7b_q4();
+        assert_eq!(mistral.name, "mistral-7b-instruct-q4");
+        assert!(mistral.size_mb > 0);
+
+        let clip = AvailableModels::clip_vit();
+        assert_eq!(clip.name, "clip-vit-base");
+    }
+
+    #[test]
+    fn test_downloader_creation() {
+        let downloader = ModelDownloader::new();
+        assert!(downloader.is_ok());
+    }
+}
diff --git a/src/ai/vision.rs b/src/ai/vision.rs
new file mode 100644
index 0000000..23981e3
--- /dev/null
+++ b/src/ai/vision.rs
@@ -0,0 +1,80 @@
+/// Vision and image analysis module
+use image::DynamicImage;
+use crate::error::{Result, AppError};
+
+/// Image analysis result
+#[derive(Debug, Clone)]
+pub struct ImageAnalysis {
+    pub ocr_text: String,
+    pub detected_objects: Vec<String>,
+    pub scene_description: String,
+    pub confidence: f32,
+}
+
+/// Image analyzer with NPU acceleration
+pub struct ImageAnalyzer {
+    ocr_enabled: bool,
+}
+
+impl ImageAnalyzer {
+    pub fn new() -> Self {
+        Self {
+            ocr_enabled: true,
+        }
+    }
+
+    /// Analyze a screenshot image
+    pub fn analyze(&self, image_data: &[u8]) -> Result<ImageAnalysis> {
+        // Decode image
+        let img = image::load_from_memory(image_data)
+            .map_err(|e| AppError::Image(format!("Failed to decode image: {}", e)))?;
+
+        // Extract text (OCR)
+        let ocr_text = self.extract_text(&img)?;
+
+        // For now, return basic analysis
+        // TODO: Add CLIP model for scene understanding
+        Ok(ImageAnalysis {
+            ocr_text,
+            detected_objects: Vec::new(),
+            scene_description: String::new(),
+            confidence: 0.8,
+        })
+    }
+
+    /// Extract text from image using Windows OCR
+    #[cfg(windows)]
+    fn extract_text(&self, _img: &DynamicImage) -> Result<String> {
+        // TODO: Integrate Windows.Media.Ocr API
+        // For now, return empty - will be implemented in next iteration
+        Ok(String::new())
+    }
+
+    #[cfg(not(windows))]
+    fn extract_text(&self, _img: &DynamicImage) -> Result<String> {
+        Ok(String::new())
+    }
+
+    /// Analyze with CLIP model
+    pub fn analyze_with_clip(&self, _image_data: &[u8], _text_query: &str) -> Result<f32> {
+        // TODO: Implement CLIP similarity
+        Ok(0.5)
+    }
+}
+
+impl Default for ImageAnalyzer {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_analyzer_creation() {
+        let analyzer = ImageAnalyzer::new();
+        assert!(analyzer.ocr_enabled);
+    }
+}
diff --git a/src/main.rs b/src/main.rs
index 0336f53..c55cb2b 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -90,6 +90,33 @@ enum Commands {
         #[arg(long, default_value = "2759")]
         port: u16,
     },
+
+    /// Manage AI models (download, list, info)
+    Models {
+        #[command(subcommand)]
+        action: ModelAction,
+    },
+}
+
+#[derive(Subcommand)]
+enum ModelAction {
+    /// List available models to download
+    List,
+
+    /// Show downloaded models
+    Downloaded,
+
+    /// Download a specific model
+    Download {
+        /// Model name (mistral, clip, minilm)
+        model: String,
+    },
+
+    /// Show model info
+    Info {
+        /// Model name
+        model: String,
+    },
 }
 
 #[tokio::main]
@@ -129,6 +156,9 @@ async fn main() -> Result<()> {
         Commands::Serve { password, port } => {
             serve_dashboard(&config, password, port).await?;
         }
+        Commands::Models { action } => {
+            handle_models_command(action)?;
+        }
     }
 
     Ok(())
@@ -308,6 +338,107 @@ fn export_data(config: &config::Config, password: &str, output: PathBuf) -> Resu
     Ok(())
 }
 
+/// Handle models command
+fn handle_models_command(action: ModelAction) -> Result<()> {
+    let downloader = ai::ModelDownloader::new()?;
+
+    match action {
+        ModelAction::List => {
+            println!("\n=== Available AI Models ===\n");
+
+            let models = vec![
+                ai::AvailableModels::mistral_7b_q4(),
+                ai::AvailableModels::clip_vit(),
+                ai::AvailableModels::minilm(),
+            ];
+
+            for model in models {
+                let downloaded = if downloader.is_downloaded(&model) {
+                    " [DOWNLOADED]"
+                } else {
+                    ""
+                };
+
+                println!("{}{}", model.name, downloaded);
+                println!("  Description: {}", model.description);
+                println!("  Size: {} MB", model.size_mb);
+                println!("  Download: activity-tracker models download {}", model.name.split('-').next().unwrap());
+                println!();
+            }
+        }
+
+        ModelAction::Downloaded => {
+            let models = downloader.list_downloaded()?;
+
+            if models.is_empty() {
+                println!("\nNo models downloaded yet.");
+                println!("Use 'activity-tracker models list' to see available models.");
+            } else {
+                println!("\n=== Downloaded Models ===\n");
+                for model in models {
+                    println!("  {}", model);
+                }
+            }
+        }
+
+        ModelAction::Download { model } => {
+            let config = match model.as_str() {
+                "mistral" => ai::AvailableModels::mistral_7b_q4(),
+                "clip" => ai::AvailableModels::clip_vit(),
+                "minilm" => ai::AvailableModels::minilm(),
+                _ => {
+                    println!("Unknown model: {}", model);
+                    println!("Available: mistral, clip, minilm");
+                    return Ok(());
+                }
+            };
+
+            match downloader.download(&config) {
+                Ok(path) => {
+                    println!("Model ready: {}", path.display());
+                }
+                Err(e) => {
+                    println!("\n{}", e);
+                }
+            }
+        }
+
+        ModelAction::Info { model } => {
+            let config = match model.as_str() {
+                "mistral" => ai::AvailableModels::mistral_7b_q4(),
+                "clip" => ai::AvailableModels::clip_vit(),
+                "minilm" => ai::AvailableModels::minilm(),
+                _ => {
+                    println!("Unknown model: {}", model);
+                    return Ok(());
+                }
+            };
+
+            let downloaded = if downloader.is_downloaded(&config) {
+                "Yes"
+            } else {
+                "No"
+            };
+
+            println!("\n=== Model Info: {} ===\n", config.name);
+            println!("Description: {}", config.description);
+            println!("Size: {} MB", config.size_mb);
+            println!("Downloaded: {}", downloaded);
+            println!("URL: {}", config.url);
+
+            if !downloader.is_downloaded(&config) {
+                println!("\nTo download:");
+                println!("  activity-tracker models download {}", model);
+            } else {
+                let path = downloader.get_model_path(&config);
+                println!("\nPath: {}", path.display());
+            }
+        }
+    }
+
+    Ok(())
+}
+
 /// Start web dashboard server
 async fn serve_dashboard(
     config: &config::Config,
diff --git a/src/web/data_export.rs b/src/web/data_export.rs
new file mode 100644
index 0000000..5228f8a
--- /dev/null
+++ b/src/web/data_export.rs
@@ -0,0 +1,126 @@
+/// Data export and retrieval endpoints
+use axum::{
+    extract::{State, Path, Query},
+    response::Json,
+    http::StatusCode,
+};
+use serde::{Deserialize, Serialize};
+use chrono::{DateTime, Utc};
+use base64::{Engine as _, engine::general_purpose};
+
+use crate::storage::Database;
+use super::state::AppState;
+
+#[derive(Debug, Deserialize)]
+pub struct ExportQuery {
+    #[serde(default)]
+    include_screenshots: bool,
+    #[serde(default)]
+    include_analysis: bool,
+}
+
+#[derive(Debug, Serialize)]
+pub struct CaptureDetail {
+    pub id: String,
+    pub timestamp: DateTime<Utc>,
+    pub window_title: String,
+    pub window_process: String,
+    pub is_active: bool,
+    pub category: Option<String>,
+    pub confidence: Option<f32>,
+    pub has_screenshot: bool,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub screenshot_base64: Option<String>,
+}
+
+#[derive(Debug, Serialize)]
+pub struct DataExport {
+    pub total_captures: usize,
+    pub captures: Vec<CaptureDetail>,
+    pub export_timestamp: DateTime<Utc>,
+}
+
+/// GET /api/captures - Get all captures (paginated)
+pub async fn get_captures(
+    State(state): State<AppState>,
+    Query(query): Query<ExportQuery>,
+) -> std::result::Result<Json<DataExport>, (StatusCode, String)> {
+    let db = Database::new(&*state.db_path, &state.password)
+        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
+
+    // Get last 30 days
+    let end = Utc::now();
+    let start = end - chrono::Duration::days(30);
+
+    let stored_captures = db.get_captures_by_date_range(start, end)
+        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
+
+    let captures: Vec<CaptureDetail> = stored_captures
+        .into_iter()
+        .map(|c| {
+            let has_screenshot = c.screenshot_data.is_some();
+            let screenshot_base64 = if query.include_screenshots {
+                c.screenshot_data.as_ref().map(|data| general_purpose::STANDARD.encode(data))
+            } else {
+                None
+            };
+
+            CaptureDetail {
+                id: c.capture_id,
+                timestamp: c.timestamp,
+                window_title: c.window_title,
+                window_process: c.window_process,
+                is_active: c.is_active,
+                category: c.category,
+                confidence: c.confidence,
+                has_screenshot,
+                screenshot_base64,
+            }
+        })
+        .collect();
+
+    Ok(Json(DataExport {
+        total_captures: captures.len(),
+        captures,
+        export_timestamp: Utc::now(),
+    }))
+}
+
+/// GET /api/captures/:id - Get specific capture with screenshot
+pub async fn get_capture_by_id(
+    State(state): State<AppState>,
+    Path(capture_id): Path<String>,
+) -> std::result::Result<Json<CaptureDetail>, (StatusCode, String)> {
+    let db = Database::new(&*state.db_path, &state.password)
+        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?;
+
+    let capture = db.get_capture(&capture_id)
+        .map_err(|e| (StatusCode::INTERNAL_SERVER_ERROR, e.to_string()))?
+        .ok_or_else(|| (StatusCode::NOT_FOUND, "Capture not found".to_string()))?;
+
+    let screenshot_base64 = capture.screenshot_data.as_ref().map(|data| general_purpose::STANDARD.encode(data));
+
+    Ok(Json(CaptureDetail {
+        id: capture.capture_id,
+        timestamp: capture.timestamp,
+        window_title: capture.window_title,
+        window_process: capture.window_process,
+        is_active: capture.is_active,
+        category: capture.category,
+        confidence: capture.confidence,
+        has_screenshot: capture.screenshot_data.is_some(),
+        screenshot_base64,
+    }))
+}
+
+/// GET /api/export/full - Export everything as JSON
+pub async fn export_full_data(
+    State(state): State<AppState>,
+) -> std::result::Result<Json<DataExport>, (StatusCode, String)> {
+    let query = ExportQuery {
+        include_screenshots: false,
+        include_analysis: true,
+    };
+
+    get_captures(State(state), Query(query)).await
+}
diff --git a/src/web/mod.rs b/src/web/mod.rs
index 140c8b9..e4c6022 100644
--- a/src/web/mod.rs
+++ b/src/web/mod.rs
@@ -2,6 +2,7 @@
 pub mod server;
 pub mod routes;
 pub mod state;
+pub mod data_export;
 
 pub use server::serve_dashboard;
 pub use state::AppState;
diff --git a/src/web/server.rs b/src/web/server.rs
index 15c5b13..3bb7e07 100644
--- a/src/web/server.rs
+++ b/src/web/server.rs
@@ -12,7 +12,7 @@ use std::net::SocketAddr;
 use std::path::PathBuf;
 
 use crate::error::Result;
-use super::{routes, state::AppState};
+use super::{routes, data_export, state::AppState};
 
 /// Start the web dashboard server
 pub async fn serve_dashboard(
@@ -33,6 +33,9 @@ pub async fn serve_dashboard(
         .route("/health", get(routes::health_check))
         .route("/stats", get(routes::get_stats))
         .route("/dashboard", get(routes::get_dashboard_data))
+        .route("/captures", get(data_export::get_captures))
+        .route("/captures/:id", get(data_export::get_capture_by_id))
+        .route("/export/full", get(data_export::export_full_data))
         .with_state(state.clone());
 
     // Build main application