by Axiom Studios

Your AI Prompt,
Evolved.

Submit your prompt. Watch two AI agents battle it out. The winner's traits survive — producing a stronger prompt every round.

We Aspire to Inspire. · Adapt. Improvise. Overcome.

+5.7
Avg score lift
3
Battle rounds
~$0.005
Cost per run
137
Tests passing
⚔️

Battle

Two prompt variants compete on your questions. They argue, rebut, and get judged. The best traits carry forward — automatically.

🧬

Improve

Not ready for a full battle? Paste any prompt and get three improved versions — sharper clarity, richer depth, stronger character — in seconds.

🏛️

Council

Ask a question. Five evolved AI minds — each shaped by battles won and lost — debate it. A synthesis agent distills it into a clear answer.

Evolution, not trial and error.

Most prompt engineering is guesswork. We apply structured selection pressure — the same mechanism that's been producing better results for 3.8 billion years.

📝
Step 1

Paste your prompt

Drop in your current system prompt and a couple of questions that represent how you actually use it.

⚔️
Step 2

Battle it out

Two variants of your prompt compete on your questions. They respond, rebut, and get scored by an impartial AI judge.

🧬
Step 3

Survival of the fittest

The winner survives. The best ideas from the loser get absorbed. A stronger version enters the next round.

Step 4

You get a better prompt

All battle learnings distilled into one clean, improved prompt. Ready to drop in wherever you need it.

⚔️ Run a Battle →

Before vs. after. Real battles.

Actual outputs from real runs — not hand-picked examples. Click a demo below to explore the results.

⚙️ For developers

Clean REST API. Two calls.

Submit and poll. No SDKs required — just standard HTTP. Integrate prompt evolution directly into your CI/CD or agent pipelines.

POST /optimize
# 1. Submit your prompt
curl -X POST /optimize \
  -H "Content-Type: application/json" \
  -d '{
  "system_prompt": "You are a support agent...",
  "test_questions": [
    "My order is 2 weeks late",
    "I want a refund now"
  ],
  "n_rounds": 3
}'

# Returns immediately:
{
  "job_id": "a3f9b2...",
  "status": "pending"
}
GET /jobs/{id}
# 2. Poll for results
curl /jobs/a3f9b2...

# When done:
{
  "status": "done",
  "result": {
    "winner_prompt": "You are a support
    agent who leads with
    empathy, then resolution...",
    "score_delta": 6.3,
    "improvement_notes": "...",
    "rounds": [...]
  }
}
OpenAPI docs → GitHub →

🧬 Prompt Improver

Not ready for a full battle? Start here. Paste any prompt and get three improved versions — sharper clarity, richer depth, and stronger character — in seconds.

📋

Paste your prompt

Any system prompt, user prompt, or instruction. Even rough drafts work.

🔍

3 angles, instantly

Clarity — Depth — Character. Each dimension targets a different aspect of what makes a prompt great.

✂️

Copy the best one

Pick whichever improved version fits your use case. One click to copy. Done.

🧬 Try the Prompt Improver →

5 free improvements per day · No account needed · Bring your own key for unlimited

🏛️ The Elder Council

Ask a question. Five AI minds — shaped by battle — debate it.

🧬
Step 1

Battle-forged agents

Your evolved agents each carry distinct traits from battles won and lost — unique voices, formed under pressure.

💬
Step 2

Each mind responds

Every council member answers your question from their own character — no consensus, no averaging, no padding.

Step 3

Synthesis

A synthesis agent distills the debate into the clearest, most actionable insight — 3–4 sentences, no fluff.

Axiom
Gen 2
3W / 1L
"I don't argue for what sounds good. I argue for what survives cross-examination. Strike the weakest link first — if the logic holds, so does the position."
Pragma
Gen 2
3W / 1L
"Every system works until the incentives change. That's where I look first. Elegant theory breaks down at the intersection with human motivation."
Logos
Gen 2
4W / 0L
"The map is not the territory. Most arguments fail because they're fighting over maps. Name the real object of disagreement and half the battle is done."
🏛️ Consult the Council →
🏛️ 1 free session / day · 5 / day on Starter · Unlimited on Pro

Start free. Scale when you're ready.

No surprises. A 3-round optimization costs fractions of a cent in API calls.

Free
$0
No card required. Try the full thing.

  • 3 battles / day
  • Haiku model only
  • Up to 10 runs / month
  • Full battle logs
  • API access
  • 5 prompt improvements / day
  • 1 council session / day
Get Started →
Pro
$29 / mo
For power users and production workloads.

  • 50 optimizations / day
  • All models including Opus
  • 5 rounds per run
  • Full API access
  • Priority support
  • Unlimited prompt improvements
  • Unlimited council sessions
Pay-as-you-go
No commitment, no subscription
$0.05 / run
No subscription. Just pay for what you use.
  • Unlimited runs
  • 5 rounds per run
  • Full API access
  • No monthly commitment