Introducing Bulb — Your AI Startup Idea Lab

You know that moment when you have a startup idea and you're not sure if it's genius or garbage? You could spend weeks doing market research, talking to advisors, and second-guessing yourself. Or you could let four AI evaluators — each with a completely different worldview — tear it apart in under two minutes.

That's Bulb.

How It Works

Submit your startup idea — name, description, target market, revenue model — and Bulb kicks off a three-phase evaluation pipeline:

Phase 1: Four Perspectives

Your idea gets evaluated simultaneously by four AI personas, each bringing a radically different lens:

• Llama 3.1 — The Visionary. The optimistic big-picture thinker. Sees the potential, the dream, the 10x opportunity. Every idea has a world-changing angle if you squint hard enough.
• Gemma 3 — The Analyst. The ruthless, data-driven VC analyst. Wants to see the TAM, the margins, the unit economics. If the numbers don't work, the dream doesn't matter.
• Qwen 3 — The Devil's Advocate. The merciless startup critic. Has seen a thousand pitch decks and knows every way a startup can fail. Finds the holes you were hoping nobody would notice.
• Claude — The Strategist. The battle-scarred mentor with pragmatic wisdom. Balances vision with execution reality. Gives you the advice you need, not the advice you want.

Phase 2: The Debate

This is where it gets interesting. After the initial evaluations, all four AIs read each other's assessments and respond — defending their positions, attacking weak arguments, and challenging assumptions. The Visionary pushes back on the Analyst's pessimism. The Devil's Advocate shreds the Visionary's optimism. The Strategist tries to find the middle ground. It's a genuine multi-perspective debate about your idea.

Phase 3: The Verdict

Finally, Grok synthesizes everything — all four initial evaluations, all the debate arguments — and delivers a comprehensive verdict. You get a score from 1 to 10, detailed scoring across ten factors (market size, competitive moat, technical feasibility, team requirements, and more), revenue projections, and a clear recommendation: invest, promising, needs work, or pass.

Try It Now — No Signup Required

Bulb is free and open to everyone. You can jump straight in as a guest — just type your name and start submitting ideas. If you have an Overdigital account, you can sign in to keep your idea history persistent across devices.

Every completed evaluation goes into the public archive where anyone can browse, read, and vote on ideas. It's part evaluation tool, part startup idea leaderboard.

Why Build This?

The startup ecosystem has a feedback gap. Getting quality, multi-perspective evaluation of an early-stage idea is either expensive (advisors, accelerators) or biased (friends and family who don't want to crush your dream). Bulb fills that gap by giving you honest, diverse AI feedback in minutes.

Is it a replacement for real market validation? Absolutely not. But it's an incredibly useful first filter. If four different AI perspectives can all find the same fatal flaw in your idea, that's probably worth knowing before you quit your job.

What's Under the Hood

Bulb runs a hybrid architecture: Llama, Gemma, and Qwen run on a local RTX 5090 via Ollama, Claude runs through the Anthropic API, and the final verdict comes from Grok via the X.AI API. Each model brings its own strengths — local models for speed and cost efficiency, cloud models for reasoning depth.

The whole pipeline is fully automated. Submit your idea and the evaluation kicks off immediately — you can watch each phase complete in real time.

Try Bulb now →