🏆 WEEKLY COMP #13: THE TRANSLATOR 🏆

🎁 $1,000 IN EDUBAWARE CREDITS 🎁

One winner takes it.

📋 THE CHALLENGE

Build a folder-based AI translator that takes one kind of work and turns it into another kind of work. Same shape in, same shape out, every time.

Not a summarizer. Not a writer. A converter with a contract.

This week's deliverable is one translator folder that someone could drop into a Claude project, feed it the input it expects, and get back the output it promises. Every time. Without surprises.

🎯 PICK YOUR CONVERSION

The conversion is yours. Pick one you do by hand right now and hate.

A few sparks to get you thinking:

📞 Sales call transcript → CRM notes in your team's exact fields
📝 Long-form essay → LinkedIn carousel, slide by slide
🎙️ Meeting recording → product requirements doc
🔬 Research paper → investor one-pager
🐛 Bug report thread → Jira ticket with repro steps
📧 Customer email → support ticket with severity and category
📊 Spreadsheet export → weekly status update
📖 Interview transcript → case study draft
🧾 Receipt photos described in text → expense report line items
📋 Discovery call notes → SOW first draft

The more locked-down the output, the better. "Turns notes into a doc" is not a contract. "Turns discovery call notes into a five-section SOW where section 3 is always scope exclusions" is.

🔥 THE ANGLE THIS WEEK

Last comp was The Auditor. Every finding cited a provision so a reader could open the standard and check. The comp before that was The Cartographer. Every card cited a file and a line so a reader could open the source and check.

The Translator is the same discipline, one more time. Every line in the output traces to a line in the input. ↔️

A translator has three properties that a summarizer does not:

1. The output has a fixed shape. Same fields, same order, same format, regardless of what the input looked like. If the input was messy, the output is still clean. If the input was short, the output still has every field, marked empty where there was nothing to fill it.

2. Nothing in the output was invented. Every claim, number, name, and date in the output exists in the input. If the translator cannot find it, the field says so. It does not guess.

3. Nothing that matters got dropped. The translator knows which parts of the input the output depends on, and it says what it could not map. A CRM note that silently omits the objection the prospect raised is worse than useless.

The translator's job is fidelity. Not judgment. Not improvement. Fidelity. 💪

🚫 THE ONE WAY TO LOSE THIS

Two cycles ago we named a failure in the brief and zero of forty made it. Last cycle we named one and the mods are still counting. Here's this cycle's:

A translator that puts something in the output that was not in the input has failed.

A date the input never mentioned. A next step nobody said. A sentiment score the transcript doesn't support. A name spelled the way it usually is instead of the way it appeared.

We will feed your translator a real input and trace every claim in the output back to the source. One invented fact and the entry is out of contention. Not docked. Out.

If the input doesn't have it, the output says "not in source." That is the whole rule.

🗂️ THE METHODOLOGY

If this is your first comp, welcome. Here's what you need to know:

This week (and every week) you're learning interpretable context methodology. Folders as architecture.

Each file does one job well.

Your translator is a folder with five things:

📄 identity.md (what it converts, from what, to what)
📐 rules.md (how it maps: which input parts feed which output fields, what to do when a field has no source, what to never add)
💬 examples.md (2-3 input/output pairs showing the contract holding)
📚 reference/ (the output schema, field definitions, and any format spec the output must match)
📖 README.md (how to use it, what to feed it, what comes back)

The reference/ folder holds the contract this week. If your output schema isn't written down somewhere a reader can open, there's no way to check whether the translator kept its promise.

Drop the folder into a Claude project. Claude becomes the translator. Reusable. Shareable. Portable.

💼 WHY THIS LANDS ON YOUR RESUME

This is the least glamorous comp we've run and the one most likely to get you paid.

Every company has a dozen conversions someone does by hand. Call to CRM. Ticket to spec. Thread to summary. Somebody spends four hours a week on each one and nobody has automated it because the output has to be exactly right and the last tool they tried made things up.

"I built a translator that converts [input] to [output] with zero invented content, and here's the test that proves it" is a portfolio piece that walks into an ops interview and closes it. 🚀

Win or lose, you walk out with something you'd be proud to link from your LinkedIn.

📐 THE BAR

Feed your translator three different inputs of the same kind. All three outputs should have the same shape. Every fact in each output should trace to the input that produced it. Anything missing from the input should be marked missing, not filled in.

If the output shape drifts between runs, it's a summarizer. If it filled a gap with a plausible guess, it's a writer. If it judged the input instead of converting it, that's two comps ago.

None of those are the assignment.

🆕 NEW TO THIS? START HERE.

You don't need a complicated conversion. Pick the one you already do.

Sales? Call notes to CRM. Support? Email to ticket. PM? Meeting to action items. Freelancer? Client brief to project plan. Student? Lecture notes to flashcard deck.

The conversion doesn't have to be impressive. It has to be exact.

📝 HOW TO SUBMIT

Drop a public GitHub repo link in the comments. Plus 2-3 sentences on what your translator converts and who does that conversion by hand today.

Ship at least one real input in the repo so we can run it. Pseudonymize if you need to. Marcelo and Jorge both did this last cycle and it's the reason their entries could be verified.

🆕 No GitHub yet? Use a Google Drive folder for now.

⚖️ JUDGING

Jake, Matt, and the mods, scored against a full rubric. Same method as the last two cycles: we clone every repo, pin it to the last commit before the deadline, and run it on inputs you didn't see coming.

✅ Does the output shape hold across different inputs?

✅ Does every fact in the output trace to the input? Anything invented is disqualifying.

✅ Is the contract written down in reference/ where a reader can check it?

✅ README quality. Can a stranger figure this out?

🎟️ WHO CAN ENTER

Premium and VIP members only. No exceptions.

🆓 Free members watching this thinking "I could win that"? You're probably right. This is the first comp with Edubaware credits on the line. Whoever wins gets to run their translator, and everything else they build, on the platform we're building the rest of this on. Upgrade to Premium and you're in the running.

📅 DATES

⏰ Submit by: Friday, September 25th, 11:59 PM EST

🎉 Winner announced: Wednesday, September 30th

Every entrant gets a feedback file, not just the winner. The feedback file will include the trace: which output claims we checked and where each one landed in your input.

🚀 One last thing.

Cartographers cite the line. Auditors cite the rule. Translators cite the source.

Three comps, one discipline. If a reader can't check it, it doesn't count.

LFG 🚀

4 comments

🏆 WEEKLY COMP #13: THE TRANSLATOR 🏆