VeriCouncil: Windows App Runs One Prompt Through Three Models, Keeps the Disagreement
Last_Aspect2229 · reddit · 2026-09-28
An indie developer launched VeriCouncil, a Windows desktop app (free for the first 100 users) born from a year of manually cross-checking ChatGPT, Claude, and Gemini answers. One prompt fans out to three providers with assigned roles — Lead drafts, Reviewer checks, Red team attacks. The app deliberately refuses to vote, average, or synthesize, since collapsing disagreement destroys signal. The author's open question for early users: when does role assignment become theater instead of signal, particularly whether Red team goes soft on confidently written answers.
More from coding & agent
- Manus ships massive update, unveils Cue agent that can join group chats and take calls — ring_hyacinth · 2026-09-28
- Claude Code Turns Rental Floor Plans into CAD, Blender Renders and Web Pages — yangyi · 2026-09-28
- Small context-structure tweaks for KV caching can slash your AI agent bills — HistoricalTerm1570 · 2026-09-28
- Turning incident outcomes into hindsight memories for AI agents — withVivek · 2026-09-28
- Google launches Credentials API to keep agent sandbox secrets off-limits — _philschmid · 2026-09-28
- Dev Pivots: Not Building a Coding Agent, But a Place to Run Your Own — lucasmeijer · 2026-09-28