# Can Ai Replace a data analyst? with Rahul Trehan on Chat GPT Auto GPT & Tableau

> This is content from just-tim, the data-and-analytics channel by Tim Ngwena (formerly 'Tableau Tim'). Tim has 12+ years of hands-on BI experience and covers Tableau most of all, plus Power BI, Looker, Hex, SQL and data modelling, the analytics industry, and the craft of doing the job — always tool-agnostic and honest about the trade-offs.

- **Author:** Tim Ngwena (just-tim, https://just-tim.com/about)
- **Published:** 2023-04-26
- **Format:** Video · 2506 min watch · transcript available
- **Topics:** AI & ML, Analytics, Data visualisation
- **Tools:** Python; SQL (joins); Tableau (ai, extensions)
- **Canonical:** https://just-tim.com/posts/can-ai-replace-a-data-analyst-conversation-with-rahul-trehan-on-chat-gpt-auto-gpt-and-tableau
- **Watch:** https://www.youtube.com/watch?v=l1CLr4BCPYc

I sit down with Rahul Trehan from Salesforce to explore whether AI can replace the data analyst. He walks me through a Tableau ChatGPT extension he built, then a more advanced AutoGPT SQL agent that translates business questions into queries, runs them against a data lake and even corrects its own mistakes.

## Key takeaways

- ChatGPT reads crosstab data as tokens rather than performing visual or true mathematical analysis, which is why numbers can fail to add up at higher 'temperature' settings.
- The system prompt assigns the model a character or role (for example, 'be a data analyst'), shaping the kind of output you get.
- AutoGPT-style SQL agents use toolkits to list tables, scan schemas, generate SQL via the LLM, fire queries against a data lake and self-correct failed queries.
- Large language models are statistical models of language, not maths engines, so expecting reliable calculations or maps from them misunderstands what they are.
- None of these AI extensions or agents are production-ready today; governance, auditing and number verification all remain unsolved work.

## Chapters

- 0:00 Why AI replaces nothing but empowers analysts
- 1:05 Rahul introduces himself
- 1:27 Why AI took off in recent months
- 3:57 From playing with ChatGPT to building an extension
- 7:56 Demo: the Tableau ChatGPT extension
- 9:32 Temperature settings and accuracy
- 11:59 Understanding the system prompt
- 13:23 LLMs as language not maths models
- 14:41 Discovering AutoGPT and SQL agents
- 17:42 The agent workflow against a data lake
- 23:15 Live coding demo of the SQL agent
- 33:37 Enterprise readiness and governance

Watch the full video, read the transcript and use chapter deep-links on the page: https://just-tim.com/posts/can-ai-replace-a-data-analyst-conversation-with-rahul-trehan-on-chat-gpt-auto-gpt-and-tableau

---
just-tim — Data and analytics, with a point of view. · https://www.youtube.com/channel/UC7HYxRWmaNlJux-X7rNLZyw · https://twitter.com/TableauTim · https://www.linkedin.com/in/timngwena