GROWTH PATTERN

Building AI Usage Tracking for LangSmith

While LangSmith offers powerful AI features, usage tracking shows 70% of users never try them without activation prompts.

LangSmith

Last updated:

What's Missing in Standard LangSmith Architecture

LangSmith provides AI capabilities, but usage tracking shows most users never try them. Without activation loops, 70% of users miss the core AI value proposition.

Standard LangSmith Flow vs Optimized Skene Flow

Standard Flow
1

User has access to LangSmith features

2

Features available but usage not monitored

3

No activation prompts for unused features

Skene Flow
1

Monitor LangSmith usage patterns and feature adoption

2

Identify users who haven't used key features

3

Trigger feature discovery emails with examples and use cases

4

Track feature adoption and send advanced usage guides

Visual comparison of the flows:

Loading diagram...

How Skene Fixes This

This Skene configuration tracks LangSmith AI API calls and identifies users who haven't used AI features. It automatically triggers discovery emails with examples and use cases to drive adoption.

Implementation Comparison

Using Skene Infrastructure

Install via Prompt

cursor.mdc
@task: Initialize Skene.
@action: Analyze my local code, validating subscription via `npx skene login`, and generate `skene.config.ts` to implement the AI Activation pattern for LangSmith.

Copy Skene Prompt for Cursor

Generated skene.config.ts

skene.config.ts
// skene.config.ts - The Automated Way
import { defineLoop } from '@skene/sdk';

export default defineLoop({
  type: 'retention',
  opinion: 'Detect dormant users and trigger personalized re-engagement campaigns',
  steps: [
    {
      trigger: {
        type: 'schedule',
        cron: '0 9 * * *' // Daily at 9 AM
      },
      condition: {
        type: 'query',
        query: `SELECT * FROM users WHERE last_activity_at < NOW() - INTERVAL '7 days' AND re_engagement_sent = false`,
        timeout: '10m'
      },
      action: {
        type: 'email',
        provider: 'resend',
        template: 're_engagement',
        personalization: {
          name: '{{user.name}}',
          lastActivity: '{{user.last_activity_at}}'
        }
      }
    }
  ],
  recovery: {
    retries: 3,
    backoff: 'exponential'
  }
});

Frequently asked questions