diff --git a/apps/app/src/instrumentation.ts b/apps/app/src/instrumentation.ts
index bf57f96d9..ce1d8d8c5 100644
--- a/apps/app/src/instrumentation.ts
+++ b/apps/app/src/instrumentation.ts
@@ -1,7 +1,12 @@
 import { diag } from '@opentelemetry/api';
+import { logs } from '@opentelemetry/api-logs';
 import { OTLPLogExporter } from '@opentelemetry/exporter-logs-otlp-http';
 import { OTLPMetricExporter } from '@opentelemetry/exporter-metrics-otlp-http';
-import { BatchLogRecordProcessor } from '@opentelemetry/sdk-logs';
+import {
+  BatchLogRecordProcessor,
+  LoggerProvider,
+  SimpleLogRecordProcessor,
+} from '@opentelemetry/sdk-logs';
 import { PeriodicExportingMetricReader } from '@opentelemetry/sdk-metrics';
 import { OTLPHttpJsonTraceExporter, registerOTel } from '@vercel/otel';
 
@@ -11,20 +16,30 @@ export function register() {
 
   const otelEndpoint = process.env.OTEL_EXPORTER_OTLP_ENDPOINT;
   const headers = parseHeaders(process.env.OTEL_EXPORTER_OTLP_HEADERS);
+  const isEdgeRuntime = process.env.NEXT_RUNTIME === 'edge';
 
-  // Configure log export
-  const logRecordProcessors =
-    // Logging doesn't currently work on edge with OTel
-    process.env.NEXT_RUNTIME !== 'edge' && otelEndpoint
-      ? [
-          new BatchLogRecordProcessor(
-            new OTLPLogExporter({
-              url: `${otelEndpoint}/v1/logs`,
-              headers,
-            }),
-          ),
-        ]
-      : undefined;
+  // Configure log export with edge runtime workaround
+  // See: https://github.com/vercel/otel/issues/104
+  // The workaround is to create our own LoggerProvider with empty logRecordLimits
+  // to avoid "Cannot read properties of undefined (reading 'attributeCountLimit')" error
+  if (otelEndpoint) {
+    const logExporter = new OTLPLogExporter({
+      url: `${otelEndpoint}/v1/logs`,
+      headers,
+    });
+
+    // Use SimpleLogRecordProcessor for edge (more compatible), BatchLogRecordProcessor for Node.js
+    const logProcessor = isEdgeRuntime
+      ? new SimpleLogRecordProcessor(logExporter)
+      : new BatchLogRecordProcessor(logExporter);
+
+    const loggerProvider = new LoggerProvider({
+      // Empty logRecordLimits is the workaround for edge runtime bug
+      logRecordLimits: {},
+      processors: [logProcessor],
+    });
+    logs.setGlobalLoggerProvider(loggerProvider);
+  }
 
   // Configure trace exporter
   const traceExporter = otelEndpoint
@@ -34,23 +49,23 @@ export function register() {
       })
     : undefined;
 
-  // Configure metrics exporter
-  const metricReaders = otelEndpoint
-    ? [
-        new PeriodicExportingMetricReader({
-          exporter: new OTLPMetricExporter({
-            url: `${otelEndpoint}/v1/metrics`,
-            headers,
+  // Configure metrics exporter (metrics not supported on edge runtime)
+  const metricReaders =
+    otelEndpoint && !isEdgeRuntime
+      ? [
+          new PeriodicExportingMetricReader({
+            exporter: new OTLPMetricExporter({
+              url: `${otelEndpoint}/v1/metrics`,
+              headers,
+            }),
+            exportIntervalMillis: 5000,
           }),
-          // Logs are also flushed on exit
-          exportIntervalMillis: 5000,
-        }),
-      ]
-    : undefined;
+        ]
+      : undefined;
 
   registerOTel({
     serviceName: process.env.OTEL_SERVICE_NAME || 'common',
-    logRecordProcessors,
+    // Don't pass logRecordProcessors - we've already set up our own LoggerProvider above
     traceExporter,
     metricReaders,
   });