Skip to content
How I AIHow I AI

“Vibe analysis”: How Faire uses Cursor, enterprise search, and custom agents to analyze data

Tim Trueman and Alexa Cerf from Faire’s data team demonstrate how AI tools are revolutionizing data analysis workflows. They show how data teams, product managers, and engineers can use tools like Cursor, ChatGPT, and custom agents to investigate business metrics, analyze experiment results, and extract insights from user surveys—all while dramatically reducing the time and technical expertise required. *What you’ll learn:* 1. How to use AI to investigate sudden drops in business metrics by searching documentation and codebases 2. Techniques for creating a semantic layer that helps AI understand your business data 3. How to build end-to-end analytics workflows using Cursor and Model Context Protocols (MCPs) 4. Ways to automate experiment analysis and create standardized reports 5. How AI can help design and analyze customer surveys 6. Strategies for creating executive-ready documents from raw data analysis 7. Why every team member should have access to code repositories—not just engineers *Brought to you by:* Zapier—The most connected AI orchestration platform: https://try.zapier.com/howiai Brex—The intelligent finance platform built for founders: https://brex.com/howiai *Where to find Tim Trueman:* LinkedIn: https://www.linkedin.com/in/tim-trueman-99788592/ *Where to find Alexa Cerf:* LinkedIn: https://www.linkedin.com/in/alexandra-cerf/ *Where to find Claire Vo:* ChatPRD: https://www.chatprd.ai/ Website: https://clairevo.com/ LinkedIn: https://www.linkedin.com/in/clairevo/ X: https://x.com/clairevo *In this episode, we cover:* (00:00) Introduction to Tim and Alexa from Faire (02:53) The challenge of analyzing product quality and usage (04:14) Breaking down what analytics actually involves beyond data manipulation (05:46) Demo: Investigating a conversion rate drop using enterprise AI search (09:05) Using ChatGPT Deep Research to analyze code changes (12:40) Leveraging Cursor as the ultimate context engine for code analysis (18:55) Analyzing a new product feature’s performance with Cursor (26:27) How semantic layers make AI tools more effective for data analysis (30:00) Using Model Context Protocols (MCPs) to connect AI with data tools (34:17) Creating visualizations and dashboards with Mode integration (37:04) Generating structured analysis documents with Notion integration (44:39) Building custom agents to automate experiment result documentation (53:10) Designing and analyzing customer surveys (59:40) Lightning round and final thoughts *Tools referenced:* • Cursor: https://cursor.com/ • ChatGPT: https://chat.openai.com/ • Notion: https://www.notion.so/ • Snowflake: https://www.snowflake.com/ • Mode: https://mode.com • Qualtrics: https://www.qualtrics.com/ • GitHub: https://github.com/ *Other references:* • Model Context Protocol (MCP): https://www.anthropic.com/news/model-context-protocol • Faire Careers: https://www.faire.com/careers _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email jordan@penname.co._

Claire VohostTim TruemanguestAlexa Cerfguest
Nov 3, 20251h 3mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Faire’s “vibe analysis” stack: Cursor, MCPs, search, agents workflows

  1. The episode reframes analytics as mostly context gathering and interpretation—not just crunching numbers—and shows how AI tools drastically shorten the “figure out what happened” phase.
  2. Tim demos enterprise AI search (via Notion) to quickly surface hypotheses for a conversion drop, then uses ChatGPT Deep Research and Cursor to forensically trace code changes tied to a checkout friction source (EORI).
  3. Alexa walks through an end-to-end feature performance analysis: pulling implementation context from the codebase, generating and executing Snowflake SQL via MCP, reviewing results in a Mode dashboard via MCP, and drafting a structured Notion doc via MCP.
  4. They close with operational automation: a custom Cursor agent that writes standardized experiment readouts from Eppo results into Notion (and a Slack summary), plus a fast workflow for designing and analyzing customer surveys using ChatGPT Projects and structured outputs from Qualtrics.

IDEAS WORTH REMEMBERING

5 ideas

Most analytics time is spent on context, not calculations.

They argue the hardest part is knowing what to ask, where data lives, and what changed—AI meaningfully improves speed and quality by making context discovery self-serve.

Enterprise AI search turns “What happened?” into a hypothesis list in minutes.

By querying Notion AI over time-bounded sources (PRDs, XP docs, launch announcements) across Slack/Notion/Jira, Tim quickly narrows likely causes of a conversion drop without manual document spelunking.

Code history is a high-fidelity source of truth for product reality.

PRDs can drift from implementation; querying GitHub via Deep Research/Cursor produces an accurate timeline of what shipped, when, and who was impacted—critical for incident-style investigations.

Cursor acts as a “context engine” when paired with MCPs.

Instead of copy/pasting across tools, Cursor can pull repo context and directly invoke connected systems (e.g., Snowflake, Mode, Notion, Eppo), reducing context switching and enabling iterative analysis loops.

Semantic layers dramatically improve AI’s zero-shot SQL accuracy.

Faire’s structured semantic definitions (business terms, joins, metrics) help LLMs map natural language to the right tables/fields, enabling faster, more reliable query generation and democratized self-serve questions.

WORDS WORTH SAVING

5 quotes

Everyone's talking about vibe coding, but no one's really talking about vibe analysis.

— Tim Trueman

The most important, often the most difficult thing, is actually just getting the right context in the first place.

— Tim Trueman

Cursor is the ultimate context engine.

— Tim Trueman

It's not the AI's name on this analysis, it's mine.

— Alexa Cerf

Don't do it in your head.

— Claire Vo

“Vibe analysis” vs traditional analytics workflowsEnterprise AI search across Notion/Slack/JiraCodebase as a data source for incident forensicsChatGPT Deep Research vs Cursor for code investigationCursor + Snowflake MCP for SQL generation and executionSemantic layers to improve AI’s data understandingMode and Notion MCPs for analysis-to-communicationCustom agents for experiment documentation (Eppo)Survey design, export cleanup, and hypothesis scoring

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome