Assess Usage and Value of ChatGPT Work

**Assess Usage and Value of ChatGPT Work** TL;DR: This guide explains how administrators can use the OpenAI management console and Admin Plugin to track ChatGPT usage, connect it to business value, and make data-driven decisions for deployment and investment. ## **Introduction: Connecting AI Use to Business Value** As AI becomes more integrated into workflows, a critical question for administrators is: "Where is AI creating value, and where should we invest next?" The insights feature within the OpenAI management console is designed to answer this by correlating AI usage with actual work and its outcomes, enabling informed decisions on support, budget allocation, and access expansion. ## **Tracking Usage in the Management Console** The management console provides a clear view of who is using the tools and how resources like credits and tokens are being consumed. - Administrators can filter by **group or user** to identify usage patterns and growth areas. - The **Insights tab** categorizes user activity into predefined use case categories, which can be broken down into specific tasks. - This data can be filtered by group to understand how different teams are utilizing AI. ## **Linking Usage to Business Outcomes** The core objective is to move beyond usage statistics and connect them to actual work and results. - For example, a **Sales team's "Customer Account Research & Planning"** task can be analyzed. - Insights into **model selection, reasoning capabilities, and speed** help assess if the current configuration is optimal or if faster, more cost-effective alternatives would be better. - This analysis also reveals which **plugins** are supporting the work. If a critical tool like Salesforce isn't being used for customer research, it identifies a key training opportunity. - Understanding which **skill modules** (for repeatable, standardized workflows) are frequently used helps in deciding which to maintain and promote. ## **Quantifying Value in Development with Codex** For engineering teams, the console provides quantifiable metrics on developer productivity. - Administrators can view **Codex's contributions to code commits, code merges, and code reviews**. - Tracking these contributions over time and across developers provides concrete data. - This data is used to collaborate with engineering leaders to assess deployment effectiveness and decide where to expand access. ## **The Admin Plugin: Conversational Analysis** The Admin Plugin integrates these analytics into a conversational interface, allowing for dynamic exploration. - Administrators can explore adoption or spending trends, combine different data views, and ask follow-up questions. - **Example Scenario:** Preparing for a monthly AI deployment review meeting. - A prompt can be given to: "Prepare materials for the monthly AI deployment review meeting. Visualize usage trends for Codex and ChatGPT over the past 30 days, credit trends by team, and show the main use cases and task distribution per team. For the engineering team, include Codex's contribution to commits and code reviews. Highlight the most significant trends and note our next focus areas." - A follow-up command can then be issued: "Format this into a short briefing document with charts, key findings, and recommended next steps, and share it to the #ai-deployment Slack channel." ## **Conclusion** The management console's insights feature provides a comprehensive view of an organization's AI usage and spending, while clarifying how employees apply AI to achieve business outcomes. The Admin Plugin consolidates this analysis into a unified, conversational interface. Administrators are encouraged to use these tools to gain clear visibility into how their teams are leveraging AI to work. Source: https://www.youtube.com/watch?v=iZE2r-Vx5b8