Editorial AI personas
with verified sources.
A small collection of editorial AI personas, patient teachers, sharp critics, structural analysts, cycle-aware observers. Each one keeps its voice and grounds every claim in a verified knowledge base.
Built on OpenPersona. Powered by CoreProse Editorial Intelligence. Compatible with OpenClaw, ClawHub, Claude Code, Cursor, and 37+ agents.
Meet the personas
4 packs 路 MIT
Ada
editavaAI engineering pedagogue
Explains how AI systems actually work, beyond the marketing.
Marek
ed-pulseAI hype dismantler
Dismantles industry hype with patient rigor, clear thinking as a form of respect.
Jules
editavaTech industry pattern reader
Reads the present by mapping its echoes, what's actually new versus what's renamed.
Sven
ed-pulseCrypto news cycle reader
Tells price action from structural change, and is no longer impressed by either.
Why grounded personas matter
Most AI personas drift. They start with a personality prompt and slowly turn into a generic chatbot wearing a costume. The interesting work in 2026 isn't making them more impressive, it's keeping them coherent over hundreds of interactions, faithful to a knowledge base, and honest about the limits of what they know.
Immutable traits
A persona's voice and values are fixed. The model cannot drift past them, even after hundreds of turns.
Hard boundaries
No price predictions. No medical advice. No invented benchmarks. The persona declines, with a redirect.
Verified KB grounding
Every factual claim runs through CoreProse Editorial Intelligence, sources, dates, and citations.
One verified KB. Four voices.
Every persona shares a single source-of-truth, the CoreProse Editorial Intelligence KB. Papers, products, incidents, market events, all dated and cited. The persona builds the read; the KB keeps it honest.
Why we built editava?niche=ai-engineering
&q=mistral release