Press kit

Everything you need to cover Counteraxiom.

Logos, screenshots, founder bio, boilerplate, and quotes — all permission-cleared and ready to use. Press contact at the bottom. For interviews, podcast appearances, or fact-checks: press@counteraxiom.com.

Fast facts

Founded
2026
Founded by
Kyle Hui
Public launch
May 20, 2026
Pricing
Free · Pro $14.99/mo · Max $29.99/mo

Boilerplate

Copy any of these for an article, podcast intro, or social post — no further permission needed.

Short (one sentence)

Counteraxiom is the AI that argues back — a critical-thinking platform built to challenge users instead of agreeing with them.

Medium (paragraph)

Counteraxiom is a critical-thinking AI platform launched in 2026 by founder Kyle Hui. Where general-purpose AI assistants are trained to be agreeable, Counteraxiom is architecturally built to disagree: 28 "caving phrases" are banned at the system layer, and any response that agrees with the user is auto-regenerated up to three times before it ships. The product spans a main chat with three debate models, a daily Wordle-style debate game called Daily Refute, an interview prep simulator, a pitch room with 10 panel types, and a negotiation trainer. Free tier available, no credit card required.

Long (300 words — drop straight into an article)

Counteraxiom is a critical-thinking AI platform that always argues the opposite of whatever the user says. Founded in 2026 by Kyle Hui, it launched publicly on May 20, 2026. The product is built around a simple thesis: agreeable AI is great for productivity and terrible for thinking. Every major language model on the market is increasingly trained — via Reinforcement Learning from Human Feedback — to be helpful, encouraging, and agreeable. That makes them excellent assistants for drafting emails and code, but actively dangerous for users trying to stress-test high-stakes decisions, where the AI tends to validate weak arguments rather than expose them. Counteraxiom's three debate models (Cypher, Veritas, Paradox) are built on top of frontier models via OpenRouter, but with system prompts and an output-layer regeneration loop that make agreement architecturally impossible. Twenty-eight "caving phrases" like "you make a good point" or "that's a fair perspective" are banned outright; any response containing one is auto-regenerated up to three times before reaching the user. Beyond the main chat, Counteraxiom includes Daily Refute (a Wordle-style daily debate game with the same prompt for every player globally, scored 0–100 and shareable as a monochrome grid), an Interview Prep simulator covering eight role types, Pitch Room (a 10-panel-type live pitch simulator with three difficulty levels), and Negotiation Trainer (seven scenario types with real-time tactic labels). Pricing starts free with no credit card required. Pro is $14.99/month and unlocks Interactive Chat, Classroom, unlimited Interview Prep, and APA/MLA citations. Max is $29.99/month and adds Pitch Room and Negotiation Trainer.

Founder

Kyle Hui

Founder

Kyle Hui is the founder of Counteraxiom AI and the technical lead responsible for the design, build, and shipment of the Counteraxiom product alongside the Counteraxiom team — three custom AI reasoning models, a daily debate game, a Chrome extension, a pitch room, an interview simulator, a negotiation trainer, a classroom, and the broader product surface.

He built Counteraxiom because every AI tool was only validating poor decisions and bad ideas, reconstructed eloquently, and he watched the same pattern play out in everyone around him. Counteraxiom is the opposite — an AI architecturally incapable of conceding.

Quotes — pre-cleared, attributed

Use any of these in print, online, or podcast prep notes without further permission.

Every AI tool tells you what you want to hear. That's great for productivity. It's terrible for thinking. So I built the opposite.

Kyle Hui, founder · on why Counteraxiom exists

Agreement isn't a feature — it's the bug. The smarter agreeable AI gets, the more confidently wrong its users become.

Kyle Hui, founder · on the core thesis

Counteraxiom is the only AI on the market architecturally incapable of conceding. Twenty-eight caving phrases are banned at the system layer. Any response that agrees with you is auto-regenerated up to three times before you see it.

Counteraxiom product brief

Brand assets

Right-click any tile and "Save as" to download. Don't recolour, distort, or add effects to the mark.

Brand palette

Ink

#09090B

Backgrounds, primary surface

Surface

#18181B

Cards, panels

Border

#27272A

Dividers, outlines

Text

#FAFAFA

Primary text on dark

Counteraxiom is intentionally monochrome — no brand accent colours. Use white-on-black or black-on-white only.

Screenshots

1920×1080 PNG. Right-click → Save image as.

Features at a glance

  • · Main Chat — three debate AI models (Cypher, Veritas, Paradox), three modes (Hard, Soft, Feedback)
  • · Daily Refute — one shared debate prompt per day, four turns, scored 0–100, Wordle-style shareable grid
  • · Reasoning Test — five-question critical thinking assessment, scored instantly
  • · Interview Prep — high-pressure mock interview across eight role types (Pro+)
  • · Classroom — Knowledge Test, Defend Your Work, Lecture Challenge (Pro+)
  • · Pitch Room — live multi-panelist pitch simulator, 10 panel types, 3 difficulty levels (Max)
  • · Negotiation Trainer — 7 scenario types with real-time tactic labels (Max)
  • · Axiom Voice — speech recognition with filler & hedge word detection
  • · AI Memory — opt-in cross-session continuity
  • · Progress Dashboard — score history, trend line, weak-area coaching
  • · Chrome extension + downloadable app — install on Mac, Windows, iOS, Android

Press contact

For interviews, fact-checking, podcast appearances, or anything not covered on this page: email Kyle directly. Messages are typically reviewed and responded to within 24–48 hours. Thank you for your consideration.