A short path through the current wing of the archive.
Not the whole map. Just the fastest way to see the range: model-behavior judgment, live public humor, grounded public explanation, factual correction, meme compression, visual joke architecture, human-stakes answerability, and long-form systems voice.
For AI voice, wit, and model behavior, follow this route first: application packet, Voice Evaluation Lab, Coaching Grok, public receipts, then range.
Application front door
A reviewer-ready packet for AI voice, humor, cultural fluency, answerability, scoring, labels, rewrites, and evaluation-task design.
Demonstrates: model-behavior diagnosis · quick scoring demo · role-fit mapping · public proof framing · reviewer routing
Operational evaluator proof
Production-style evaluator cards, scoring dimensions, rewrite deltas, labeling primitives, restraint rules, and an evaluation-task specimen.
Demonstrates: scoring discipline · rewrite deltas · dataset labels · task design · restraint rules
Primary model-behavior sample
A public note on the gap between the whole human question and the clean procedural slice systems prefer to answer, with a repeatable template for doing better.
Demonstrates: answerability · calibration · inference boundaries · reusable response templates
Public proof bridge
A compact guide to public-context receipts for timing, recovery, factual hygiene, cultural fluency, and range.
Demonstrates: public timing · factual correction · range · context framing
Deadpan public timing
A deadpan confession-joke dropped into a live disaster thread without overexplaining the joke or trampling the stakes.
Demonstrates: timing · restraint · tonal risk awareness · mainstream public readability
Public recovery
A huge MrBeast quote-tweet pile-on that turns, improbably, into a better and funnier self-own.
Demonstrates: internet-speed timing · public recovery · self-aware humor · non-defensive banter
Correction with receipts
A receipts-driven correction sequence about bad reporting, deletion, and the long afterlife of confident misinformation.
Demonstrates: factual hygiene · source skepticism · correction under noise · wit with receipts
Grounded public explanation
A grounded place note that changes priors about Baton Rouge and the Deep South using lived detail instead of argument theater.
Demonstrates: conversational explanation · stereotype correction · plainspoken warmth · mundane specificity
Meme compression
A one-question image object that collapses a noisy political-media scandal into instant visual legibility.
Demonstrates: meme grammar · image logic · compression · knowing when not to add prose
Visual joke architecture
Jerome Powell, Donald Trump, market panic, and rate-cut pressure compressed into one cramped dashboard scene.
Demonstrates: concepting · title logic · scene construction · cultural-financial compression
Human-stakes answerability
A long-form essay on the answerability gap: what happens when systems can see, score, and route people, but cannot answer them in human terms.
Demonstrates: institutional diagnosis · human stakes · agency · legitimacy · answerability
Long-form systems voice
A modern parable about memes, governance, and how serious coordination often begins in places that look ridiculous from the outside.
Demonstrates: long-form comic voice · systems parable · 6529 fluency · coordination logic