diff --git a/shared/common/src/lib.rs b/shared/common/src/lib.rs
index 117c161..ed130cc 100644
--- a/shared/common/src/lib.rs
+++ b/shared/common/src/lib.rs
@@ -5,3 +5,4 @@ pub mod models;
 pub mod utils;
 pub mod webhooks;
 pub mod health;
+pub mod token_limits;
diff --git a/shared/common/src/token_limits.rs b/shared/common/src/token_limits.rs
new file mode 100644
index 0000000..24ec8ab
--- /dev/null
+++ b/shared/common/src/token_limits.rs
@@ -0,0 +1,95 @@
+//! AI Token Limit Utilities
+//! Provides functions to check and enforce monthly token limits
+
+use sqlx::{PgPool, FromRow};
+use uuid::Uuid;
+use chrono::{DateTime, Utc};
+
+#[derive(Debug, Clone, serde::Serialize, serde::Deserialize, FromRow)]
+pub struct TokenLimitCheck {
+    pub has_available_tokens: bool,
+    pub monthly_limit: i32,
+    pub used_tokens: i64,
+    pub remaining_tokens: i64,
+    pub reset_date: DateTime<Utc>,
+}
+
+#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
+pub struct TokenLimitError {
+    pub error: String,
+    pub monthly_limit: i32,
+    pub used_tokens: i64,
+    pub remaining_tokens: i64,
+    pub reset_date: DateTime<Utc>,
+}
+
+/// Verify if user has available tokens for AI operations
+/// 
+/// # Arguments
+/// * `pool` - Database connection pool
+/// * `user_id` - User UUID
+/// * `estimated_tokens` - Estimated tokens for this operation (default: 1000)
+/// 
+/// # Returns
+/// * `Ok(TokenLimitCheck)` - User has available tokens
+/// * `Err(TokenLimitError)` - User exceeded limit
+pub async fn check_ai_token_limit(
+    pool: &PgPool,
+    user_id: Uuid,
+    estimated_tokens: i32,
+) -> Result<TokenLimitCheck, TokenLimitError> {
+    let result: TokenLimitCheck = sqlx::query_as(
+        "SELECT * FROM check_token_limit($1, $2)"
+    )
+    .bind(user_id)
+    .bind(estimated_tokens)
+    .fetch_one(pool)
+    .await
+    .map_err(|e| {
+        tracing::error!("Failed to check token limit: {}", e);
+        TokenLimitError {
+            error: "Failed to verify token limit".to_string(),
+            monthly_limit: 0,
+            used_tokens: 0,
+            remaining_tokens: 0,
+            reset_date: Utc::now(),
+        }
+    })?;
+
+    if !result.has_available_tokens {
+        tracing::warn!(
+            "User {} exceeded token limit: {}/{} (reset: {})",
+            user_id,
+            result.used_tokens,
+            result.monthly_limit,
+            result.reset_date
+        );
+        
+        return Err(TokenLimitError {
+            error: format!(
+                "Monthly AI token limit exceeded. Used: {} / Limit: {}. Reset date: {}",
+                result.used_tokens,
+                result.monthly_limit,
+                result.reset_date.format("%Y-%m-%d %H:%M UTC")
+            ),
+            monthly_limit: result.monthly_limit,
+            used_tokens: result.used_tokens,
+            remaining_tokens: result.remaining_tokens,
+            reset_date: result.reset_date,
+        });
+    }
+
+    Ok(result)
+}
+
+/// Get formatted token usage message for API responses
+pub fn format_token_limit_message(used: i64, limit: i32, reset: DateTime<Utc>) -> String {
+    let percentage = (used as f64 / limit as f64 * 100.0).round();
+    format!(
+        "🚫 Límite de tokens IA alcanzado: {}% usado ({}/{} tokens). Reset: {}",
+        percentage,
+        used,
+        limit,
+        reset.format("%d/%m/%Y a las %H:%M")
+    )
+}