Spreadsheets &amp; Data Work

AI for spreadsheets: cleaning, analyzing, and charting data

A practical guide to using AI to handle spreadsheets — cleaning messy data, writing formulas, building pivot tables, and turning rows of numbers into clear answers. Works in Excel, Google Sheets, or via ChatGPT directly.

10 min read

Evals for non-engineers: know if your AI workflow is getting better or worse

Evals — systematic measurement of AI output quality — are usually treated as an engineering concern. But every team running AI workflows needs them, and the basics are accessible without code. The how-to.

Intermediate

More in this topic

7 min read

Household Budget Scenarios: What-If Math Without Financial Advice

Build transparent what-if scenarios from your own verified household figures — a job loss, a rent increase, a new baby — and see the arithmetic clearly. No investment tips, no debt strategy, no numbers that did not come from your own documents.

6 min read

Finding Subscription Drift Without Exposing Your Bank History

Review a locally redacted transaction export for recurring charges and drift, without ever connecting an AI tool to your bank account. A confidence-flagged merchant list, and a cancellation checklist a human actually executes.

56 minutes

AI Agents in Finance with HPE's Chief Financial Officer (CFO)

CXOTalk. HPE CFO Marie Myers describes rolling agentic AI (their "Alfred" platform) across a 3,600-person finance organization with the discipline the article argues for: a direct-versus-indirect ROI framework, stage gates that decide whether each use case continues or stops, mandatory human-in-the-loop quality controls, and workflow-level metrics such as a weekly review process losing roughly 90% of its manual effort. Adoption is measured as changed work, not tool logins.

9 min read

AI ROI and maturity: how to measure adoption that actually works

AI adoption should not be measured by how many people tried ChatGPT. A practical framework for measuring workflow ROI, quality, risk, maturity, and scale-readiness.

13 min read

Structured outputs and function calling: the production patterns

Structured outputs and function calling are the bridge from 'LLM that generates text' to 'system that does work'. In production, the patterns that matter are about schemas, error handling, idempotency, and graceful degradation — not just JSON mode.

18 minutes

How to Improve Your Excel Skills with ChatGPT

Leila Gharani. Older (early 2023) but still the cleanest worked example of pasting a real Excel question into ChatGPT, getting a SUMPRODUCT-or-conditional-formatting answer back, and then iterating with the model when the first formula doesn't quite fit. The model has improved a lot; the prompting workflow she shows is the same one you'll use today.

12 minutes

Excel's New AI Function is Absolutely Insane (Copilot Function)

Leila Gharani. Live demo of the new `=COPILOT()` function on the kinds of messy real data the article describes — shift handover notes turned into a structured action table, free-text reviews tagged for sentiment, and inconsistent job titles normalized against a master list. Also calls out the limits clearly: don't use it for anything that has to be exact, freeze results as values when you're done.

107 minutes

Why AI evals are the hottest new skill for product builders | Hamel Husain & Shreya Shankar

Lenny's Podcast. Hamel Husain and Shreya Shankar walk through the entire eval workflow on a real property-management AI assistant — looking at traces, open and axial coding of errors, deciding when to stop, building an LLM-as-judge, and validating it against human judgment. This is the rare long-form conversation that is genuinely aimed at PMs and team leads rather than ML engineers, and it covers the same "30 minutes a week after setup" rhythm the article recommends.

Intermediate

3 minutes

Evaluate prompts in the Anthropic Console

Anthropic. A three-minute Anthropic walkthrough of running a real eval inside the Workbench — auto-generating realistic test cases, grading outputs, tweaking the prompt, and re-running the same suite side-by-side. The view count sits below the usual bar, but for "how do I actually do this without writing code" this is the cleanest official demo and slots neatly under the more strategic Husain/Shankar conversation.

Intermediate

41 minutes

OpenAI DevDay 2024 | Structured outputs for reliable applications

OpenAI. Walks through `strict: true`, the difference from old JSON mode, refusal handling, and how function calling and response-format schemas compose. Useful precisely because it describes the contract the API gives you, which is what the article's production patterns are built on top of.

18 minutes

Pydantic is all you need: Jason Liu

AI Engineer. The talk that crystallised the modern "define a Pydantic model, hand it to the LLM, let validation do the rest" pattern, with concrete examples of nested objects, validators that catch hallucinated URLs, and Chain-of-Thought as a typed field. Watch it before re-reading the article's section on validators and you will recognise where its retry and refusal rules come from.