Useful work. Visible limits.
Independent examples and prototypes that let you look at the process, including the parts that still need a person. Each card says what kind of work it is and what you leave with.
Practical AI learningBeginner lessons and optional practice materials: Start With AI — Stay Human →
Make everyday work clearer
Start with a complete example. Keep a result you can use.
Working example · Prepared sample · Optional AI
A useful weekly update
For a team lead with scattered notes. See a finished example, then make an update that separates progress, attention, follow-up and unknowns.
You leave with: an editable update you can copy or save as a real PDF.
Working example · Prepared sample · Optional AI
Five customer replies
For a small team answering the same questions. Supply five questions and approved facts, then refine one answer without changing the others.
You leave with: five separate replies and a current PDF guide for your team.
Interactive example · Dated public references
Visitor Answer Desk
For staff answering visitor questions. Each fictional question has a prepared reply beside its dated source; edit it, review it, copy it.
You leave with: a checked reply, or a clear flag that something needs confirming.
Check information before you act
Separate what the record shows from what still needs a decision.
Working example · Prepared sample · Optional AI
Find the story behind sales
For a manager comparing two years. Download the real Excel example, change it, then compare sales and gross margin before asking AI for an interpretation.
You leave with: a calculated comparison, chart, three priorities, checks and a current PDF.
Working tool · Public and fictional data · Optional AI
Before the Import
For someone handed a spreadsheet of organizations to bring into a CRM. See which records need a decision, then draft questions for the person who made the file.
You leave with: reviewed and held records, a decision log and a note for the file owner.
Review tool · Practice exercise · Optional AI
Before You Publish the AI Draft
For a writer or editor checking a short draft. Compare it with the sources you provide, with optional AI or by hand, or practice on a fictional homepage.
You leave with: findings you've decided on, plus a record of what's still unresolved.
Interactive example · Dated EIA figure · Optional AI
Source to Story
For a writer, editor or communications lead adapting one sourced fact for a newsletter and a social post. Set your angle, ask AI for two suggestions if you want them, and review each version before handing it over.
You leave with: two drafts you can copy separately, plus a writing pack or a full review record.
Practice a decision
A short exercise, a useful comparison, or a checklist to take with you.
Interactive checklist · Dated safety sources
Ready for the Fixer
For someone taking an item to a free community repair event. Describe it at home and run two checks: battery warning signs and recalls.
You leave with: a ticket to hand the fixer, or a clear reason to wait.
Data explainer · NOAA climate normals
Which Place Is Wetter?
For anyone asking whether Seattle or New York is wetter. Compare how much falls and how often, year and month, with NOAA's 1991–2020 normals.
You leave with: a chart, the full table and a summary you can copy or print.
Training exercise · City of Portland rules
The Shift-Change Sort
For a café trainer. Sort seven items using Portland's published business compost rules, including one the City's pages don't settle.
You leave with: a station checklist and the questions to confirm.
Training scenario · Fictional
The Next Right Step
For customer-service staff. A customer expects her order on Friday and nothing confirms it: make two decisions and see what each sets off.
You leave with: an eight-step procedure and the handoff note your path produced.
Prepared examples and local calculations are labeled. Optional AI runs only after you request it and confirm the tool’s notice. Source to Story sends the evidence card, source passage, audience and point you confirm. See each tool’s data notice and the Privacy Policy. For clearer messages, visit Writing; for a small promise-checking exercise, visit Learn.
Writing samples and case studies
Long-form work to read, with sources and limits kept visible.
Technical writing sample
Research to Readable
The FinanceQA benchmark paper (January 2025) turned into an explainer a general reader can follow, with LinkedIn and X adaptations.
Inside: the explainer, two adaptations and numbered references.
Technical writing for developers · Tested code
Reading OCR Confidence
How to flag OCR values for review with Tesseract's confidence scores, while keeping the original text and its position visible.
Inside: an evidence-to-decision table and a small tested JavaScript rule.
Case study · Self-directed design and build
This studio website
Before-and-after captures of this site and the decisions behind the rebuild. Self-directed, not a client commission.
Inside: what changed, how it was directed and its stated limits.
Prototype case study · Recorded demo
Building Calqen
How a wardrobe-decision prototype was bounded during OpenAI Build Week 2026: three outfits from clothes you supply, and one targeted change.
Inside: the recorded demo, the product decisions and the build's limits.
A task of your own?
The One-Workflow Sprint begins with one recurring task and a safe example of the result you need.