AIONIS Delfos

• 

• 

  landing

• 

  idea

• 

• 

• 

• 

  novus proof

Inspiration

Every idea validator I tried did the same thing: I described an idea, and it told me the idea was great. Big market, strong trends, go for it. But it never asked the only question that actually decides whether I should build something: am I the right person to build this? A perfect idea in the wrong hands is still the wrong idea. I wanted a tool brave enough to tell me the truth I was avoiding.

What it does

Delfos is a founder–idea fit oracle that never flatters you. You describe an idea, it asks you a few sharp, idea-specific questions about yourself, and it delivers one honest verdict with two faces, both a win: RELEASE (this isn't yours; let it go and spend your energy where you're genuinely strong) or IGNITE (you're not the one today, you lack one concrete thing, but earn it and this becomes yours). It doesn't grade your idea. It weighs you against what the idea truly demands, and anchors every line of the verdict in your own words.

How we built it

Next.js with the LLM running entirely server-side behind route handlers, so the verdict logic and prompts never reach the browser. Two structured calls: one generates the guided, idea-specific questions; one emits the final verdict as validated JSON (verdict face, the decisive crux, the anchored reasoning, and the concrete challenge or release). Claude Sonnet runs both, with dual-provider failover so a live demo never dies mid-verdict.

Challenges we ran into

The hard part, like always, wasn't the code, it was the calibration of the soul. A tool that always says "you can do it if you try!" is just another flatterer wearing a brave mask. The real work was forcing Delfos to be genuinely willing to land on RELEASE when the evidence points there, without ever becoming cruel or inventing weakness the person didn't reveal. Honest, never cruel: the truth should land like a hand on the shoulder, not a slap.

Accomplishments that we're proud of

I tested it on my own ideas, and it passed the hardest test: it told me the truth. I fed it a dream-capture project I love, and it gave me RELEASE, anchored in things I'd actually said. I fed it an e-commerce tool I'd genuinely built for real paying clients, and it gave me IGNITE with a specific, uncomfortable challenge. Both verdicts were right. A tool that only flatters its own creator would be worthless. This one didn't.

What we learned

That the courage to say no is the entire product. Anyone can build a validator that says yes. The value, and the difficulty, is in a verdict that's willing to disappoint you when disappointing you is the honest thing to do, and to do it in a way that leaves you clearer and more alive, not smaller.

What's next for AIONIS Delfos

Right now Delfos delivers the verdict and stops. The vision is what comes after IGNITE: if it tells you that you lack one concrete thing, the next step is helping you go earn it, turning the verdict into the first step of becoming the founder the idea needs.

Built With

• anthropic
• claude
• next.js
• openrouter
• react
• typescript