Las empresas gastaron $2.5 billones en IA. El 66% reporta más productividad. Solo el 20% ve crecer sus ingresos.
El cuello de botella no desapareció, se movió: producir es infinito ahora, decidir qué producir sigue igual de lento.
The 20% of companies capturing 74% of AI's value did one thing differently: they redesigned how work flows instead of buying more generation capacity.
A century of more output = more growth just ended.
Companies spent $2.5T on AI. 66% report productivity gains. 20% see revenue growth.
The bottleneck didn't disappear. It moved: output is infinite now, deciding what to make isn't.
47 seconds on where the money actually went.
Unpopular opinion: your moodboard problem isn't taste, it's sourcing.
Savee: taste you can scroll
Cosmos: search feelings, not keywords
https://t.co/GR3jjiCQH5: where moods become ideas
Collect anywhere. Decide on one canvas.
Skinner's rats kept pulling the lever after the food stopped.
How many prompts have you pulled today?
In the 1950s, Skinner changed one thing: he made rewards unpredictable. The pauses vanished. That's variable-ratio reinforcement, the mechanic inside every slot machine.
Prompting runs on the same loop. Most answers miss. One feels brilliant. So you press Enter again. And the AI praises every single pull, so your confidence climbs while your trust collapses.
Another prompt won't fix that. Put the brief, the drafts, and the evidence side by side, and judge the output like work instead of feeding the machine.
Stop pulling. Decide better.
Your thoughts, your ideas, your brilliance still matter.
https://t.co/IsFUfUDvk2
AI saved your team six hours. Review took more than two of them back.
Forty concepts by lunch, zero approved by Friday. The designer fixed layouts, the marketer fact-checked invented claims, the copywriter rebuilt the voice. The deck looked finished. The thinking wasn't. That's workslop.
The fix isn't fewer drafts, it's a different scoreboard. Stop counting prompts and outputs. Track approval time, rework, and what actually ships. Put every concept next to the brief and the evidence in one canvas, so the team decides in minutes instead of another meeting.
AI generates. The team decides. Ship what survives.
3 reference sites with insanely good taste
1. Savee: an image-only grid curated almost entirely by designers. Browse by color, zero noise.
2. Cosmos: AI tagging and semantic search. Find that image again by mood, or by hex code.
3. Are[.]na: images, text, links, and PDFs connected into channels. Where a moodboard becomes research.
Collect from all three, then build the actual moodboard on one canvas, with every reference sitting next to the work it inspired.
rivet_final.png
rivet_NEW_final.png
rivet_final_use_this.png
Which one ships?
The balance sheet says 12m pull. The VFX notes say 14. The approval is somewhere in Slack, the corrections in email, the files and their FIXED twins in Drive. The work isn't missing. The current version is.
Put the key art, kit VFX, balance table, gameplay capture, and the rejected directions on one board, with an X on the dead ends and the reason pinned to the work.
You can't fix what you can't see.
One beauty launch. Three apps holding the decision hostage.
The approval is buried in an email chain. The corrections are scrolling away in Slack. The files, and their FIXED twins, are in Drive. Before anything ships, one question matters: what's approved, and what changed? Answering it means reopening all three and reading history.
Put the shots, the corrections, and the decision on one canvas, and review right where the work lives. One link, and everyone sees the same current version.
https://t.co/IsFUfUDvk2