Tool result caps need explicit output contracts
A capped search can be mistaken for a complete result. MCP structured results and JSON Schema required fields make output boundaries inspectable.
The BIG CHANGE archive
News, analysis and practical guides.
Explore our reporting on how AI and robotics change work, business and the world.
The latest change
849 stories
Humanoid safety filters failed in selected physical tests
Seven selected filter trials produced collisions; broader search totals conflict.
Make a checked competitive and market report with Gemini Deep Research
Compare named companies on the same dimensions, reconcile dated market evidence, inspect the exported Gemini report and approve an exact revision.
Chain Claude Code skills into a reviewed marketing and operations system
Build a two-Skill local workflow with separate marketing and incident sources, exact human review, a checked draft outbox, latest-run health and reviewed learnings.
Map a codebase with bounded Claude Code or Codex subagents
Map an approved repository slice with bounded Claude Code or Codex subagents, then verify tracked file hashes, exact source lines, symbols and direct dependencies by hand.
Draft five checked social posts with Claude Code and schedule them through Blotato
Use an approved source to draft five posts with Claude Code, approve exact account and time payloads, schedule through Blotato, then verify the queue and later results.
Turn an owned YouTube video into checked social posts and visual drafts
Use corrected captions from an owned YouTube video to draft source-bound LinkedIn, Instagram and X posts, plan optional original images and record a human acceptance decision for each pair.
Move checked ChatGPT context to Gemini, then verify what arrived
Inventory owned ChatGPT instructions, memories, GPTs, Projects, files and chats; use eligible Gemini imports and rebuild selected context, then check each result against the source.
Research one topic, approve a linked digest and post it to one Slack channel
Use a complete Python 3.12 recipe to search one public topic, check a cited digest, authorize one Slack channel post and inspect its exact bot readback or uncertain-delivery journal.
Analyze one public web page and check its summary, points and tone
Capture one permitted public HTML page, generate a short summary, quoted key points and estimated tone, then compare every claim with saved text before recording human approval or a hold.
Research a public topic and make a checked, linked PDF report
Use live OpenAI web search to draft a four-section report, export a complete cited PDF, correct claims locally and approve its exact bytes after human review.
Find a company and check a cited competitor profile
Find a company's official product page, keep a real retrieval receipt, draft a cited four-section profile and classify every claim before making a checked report.
Write a sourced blog draft with live web search and human review
Use a complete Node.js project to search the live web, draft a cited blog post for a chosen audience, then open the sources and resolve its claims before use.
Turn a source-checked audio brief into a WAV file
Run a cited OpenAI web search, check its sources, approve two or three spoken sentences, then generate and listen through a Gemini TTS WAV with the complete Python script.
Compare Midjourney V8.2 images from one approved brief
Use one approved brief and fixed V8.2 settings to compare two stylize values, inspect real outputs, then reopen and record a chosen image with costs and permissions.
Draft and check three outreach emails from one company page
Prepare three unsent emails from one company page and approved offer; check claims and the recipient before saving.
Check what the Economic Index says about your occupation
Match duties to an Index occupation; check cited tasks and limits.
Check a shared Stitch-to-AI-Studio app
Export Stitch screens; check the filter and viewer link.