Watch Me Do a Data Analyst Project in Minutes With Claude Code
A working data analyst does an entire real analysis project live with Claude, the fastest way to see why memorizing formulas is no longer the bottleneck.
Open youtube.com →3 questions founders actually ask, each with a straight answer and the resources worth your time.
Because you can now just ask questions in plain English, "which customers churned last quarter and why?", and tools like ChatGPT and Claude write the formulas or code behind the scenes, in minutes instead of the months it takes to get good at Excel or SQL. As a founder, your scarce resource is time, and the insight matters more than the skill of producing it. The one caveat: AI can be confidently wrong, so you still want enough judgment to sanity-check what it tells you before betting the company on it.
A working data analyst does an entire real analysis project live with Claude, the fastest way to see why memorizing formulas is no longer the bottleneck.
Open youtube.com →Written by someone who built data stacks for several startups, aimed at non-technical execs: what to upload, what to ask, and where ChatGPT's answers stop being trustworthy.
Open definite.app →The honest counterweight: AI drafts queries brilliantly but misses logically-wrong joins and filters, so learn just enough to verify, you own the results.
Open mavenanalytics.io →The basic workflow is: export your data (a CSV from Stripe, your analytics tool, or a survey), upload it to ChatGPT or Claude, and ask questions in plain English, the AI summarizes, spots trends, builds charts, and pulls themes out of hundreds of open-ended survey answers in minutes. For messier or recurring jobs, tools like Claude Code can read whole folders of files and produce full reports and dashboards. The key skills are asking specific questions, giving the AI context about your business, and spot-checking a few numbers by hand.
A free, university-quality walkthrough of uploading CSVs to ChatGPT with the exact prompts to use, copy the workflow directly onto your own sales or usage data.
Open ai-analytics.wharton.upenn.edu →A non-programmer's first-person account of getting charts, stats, and a full written report out of a raw dataset using only plain-English instructions (with free links to the full pieces).
Open technofile.substack.com →Copy-paste prompt templates for pulling themes and sentiment out of open-ended survey responses, plus an honest take on where ChatGPT breaks down past ~500 responses.
Open zonkafeedback.com →A respected founder/PM shows his real workflows, asking data questions in plain English against a database and synthesizing customer interviews (intro sections are free).
Open sachinrekhi.com →These tools sit inside your spreadsheet and work on it directly: Claude for Excel can clean messy data, write and debug formulas, explain how an inherited model works, and build multi-tab financial models with charts and dashboards from a plain-English prompt. In Google Sheets, Gemini can analyze your data conversationally, and the =AI() function lets you categorize, summarize, or generate text across a whole column. They're genuinely useful for 80% of founder spreadsheet work, but always review the output before it goes to an investor, because they can still make mistakes.
An honest stress-test on real tasks, data cleaning, formula debugging, a three-statement financial model, dashboards, so you see what it nails and where it stumbles.
Open youtube.com →First-person build of a full financial model with Claude for Excel, including the actual prompts, the closest thing to a founder playbook for this tool.
Open natesnewsletter.substack.com →The official quick intro to typing =AI() in any Sheets cell to categorize feedback, summarize reviews, or generate copy across a whole range, no add-ons needed.
Open blog.google →The straight-from-the-source list of what it can do (cell-level citations, model debugging) and can't (macros, VBA), plus security cautions worth knowing before you rely on it.
Open support.claude.com →