SMITH

Most AI gives you ideas.
We give you survivors.

12 ideas enter a pipeline of 15 AI agents and 6 kill gates. 1 survives — with a full venture memo and 48-hour build plan.

Generate Your First Idea Free

1 free credit on signup · no card required

15

AI agents

7

frameworks

92%

kill rate

10

score criteria

What you're used to vs. what this actually does

"Here are 10 startup ideas for 2026" — congratulations, you now have 10 ideas and zero conviction

Generates 12, kills 8 in sequential filters, validates the survivor with simulated buyers and modeled unit economics, then writes the full venture memo

"Build a platform that connects X with Y" — a sentence that could describe anything and helps with nothing

Every surviving idea ships with a named competitive landscape, specific GTM channels, a "Google Builds This" moat test, and a 48-hour build plan

"This is a great opportunity in a growing market" — AI-generated confidence with zero earned credibility

Every claim is backed: Tech feasibility ≥7/10, at least 2/3 simulated personas must buy, LTV:CAC modeled at >3:1, and any score below threshold kills the idea

The Forge

6 phases. 15 agents. 1 survivor.

Each phase has specialized AI agents. The output of each feeds the next. Ideas that can't survive don't.

01

Deep Market Research

5 research agents

Value-chain mapping · Pain-point archaeology · "Why 5 Times" root-cause analysis · Trend triangulation · Adjacent-market scouting

02

7-Framework Ideation

3 thinking agents

Paul Graham · YC Seven Recipes · Blue Ocean Strategy · Schlep Filter · First Principles · Contrarian Thinking · MVT Design

03

6 Kill Gates

>60% die here

"So What?" Test · Competition Obituary · "Would I Use This Tomorrow?" · GTM Clarity · "Build It In 48 Hours" · Autonomous Future Test

04

4-Way Validation

4 specialist agents

Technical Feasibility ≥7/10 · Customer Simulation (3 personas, ≥2 must buy) · Moat Architecture · Unit Economics (LTV:CAC >3:1)

05

10-Criterion Scoring

Per-criterion kill thresholds

Category Creation · AI-Native Moat · Forward Vision · Buildability · Revenue · GTM Precision · Customer Pull · Scalability · Risk · Timing

06

Full Venture Memo

2 builder agents

Elevator pitch · Business plan · Unit economics · GTM playbook · Competitive landscape · Moat roadmap · 48-hour MVP build plan

12 raw ideas generated100%
After Kill Gates~40% survive
After 4-Way Validation~15% survive
After Scoring Thresholds1 survivor

kill rate: >90% · survivor: battle-tested

What Ships With Every Survivor

Not a paragraph. An entire thesis on why this business should exist.

Pain-Point Archaeology
Digs for problems people have accepted as permanent — the schleps, the workarounds, the "we've always done it this way." That's where real businesses hide.
7 Frameworks, Not 1 Prompt
Paul Graham, YC, Blue Ocean, Schlep Filter, First Principles, Contrarian, MVT — applied systematically. Ideas that emerge from 3+ frameworks independently get flagged as strong signal.
Kill Gates That Actually Kill
6 sequential filters: "So What?", Competition Obituary, "Would I Use This Tomorrow?", GTM Clarity, 48-Hour Build Test, Autonomous Future. Most ideas don't make it. That's the point.

Moat Architecture

Data flywheels, network effects, counter-positioning, switching costs — the AI maps a defensibility roadmap because "first mover advantage" isn't a moat.

Simulated Customer Panel

3 buyer personas with specific titles, company sizes, and daily workflows. They're interrogated. At least 2 must say variants of "I need this" or the idea dies.

Unit Economics, Not Vibes

CAC, LTV, gross margin, payback period — modeled at Month 6, 12, and 24. If LTV:CAC doesn't clear 3:1, the idea doesn't survive to the memo.

Pipeline Activity Feed

See what the AI is working on in real-time: which phase is active, how many ideas were killed, when scoring begins. A transparent window into the forge.

48-Hour Build Plan

Tech stack chosen, directory structure mapped, week-1 action items listed — designed so you (or your AI coding agent) can start building the day you read it.

Score, Triage, Compare

10-criterion radar chart, drag-and-drop kanban for Hot/Warm/Cold triage, and per-criterion justifications so you know exactly why each score is what it is.

The Price of Knowing

A bad idea costs you months. A good one changes your life. We charge less than lunch.

Starter
$5

5 ideas

$1.00/idea

Most popular
Popular
$20

25 ideas

$0.80/idea

Save 20%
Pro
$60

100 ideas

$0.60/idea

Save 40%

1 free credit on signup · no credit card required

You Know Who You Are

You don't need another brainstorm. You need conviction.

The solo founder at 2 AM

You can build anything with AI tools. The hard part isn't coding — it's knowing what's worth coding. You need one idea you'd bet a year on.

The technical co-founder without a pitch

You have the skills. You need an idea backed by real research, real numbers, and a plan that makes VCs lean forward.

The serial shipper

You've launched before. Some worked, most didn't. You're done guessing. You want the guess removed from the equation.

The AI-native builder

You use Cursor, Lovable, Replit. You build in days, not months. You need ideas architected for that speed.

The worst startup mistake
is building the wrong thing.

30 seconds from now, you could have an idea that survived 15 AI agents, 6 kill gates, and a simulated customer panel. Or you could keep scrolling.

Generate Your First Idea Free

your first idea is free