Validate Before You Build

Find out if your idea deserves the next month before you build it.

A practical, evidence-first validation system for solo builders using AI, code, SaaS and digital products.

Find the assumption most likely to kill your idea, test it cheaply, and make a clear BUILD / TEST MORE / PARK / KILL decision.

Get the Founding Version — $39See How It Works

One-time founding price · File delivered within 48 hours · 7-day refund

Validation FileGround rules
AI-generated claim
Not evidence
Unsupported claim
UNKNOWN
Riskiest assumption
Tested first
Output
BUILDTEST MOREPARKKILL

The problem

Before you build, answer one harder question.

AI makes it easier than ever to build something.

That doesn't make it easier to know whether you should.

  • 01You can spend weeks researching.
  • 02You can ask ChatGPT for a market analysis.
  • 03You can build an MVP in days.
  • 04You can collect dozens of opinions.

And still not know whether the idea deserves another month of your time.

The problem isn't a lack of information.

It's knowing what evidence is actually strong enough to change your decision.

The core method

Turn an idea into a decision.

Five stages, in order. Each one narrows what you believe down to what you can defend.

  1. 01

    Assumptions

    What must be true for this idea to work?

  2. 02

    Evidence

    What evidence do you actually have?

    Where did it come from?

    How strong is it?

  3. 03

    Falsifier

    What result would make you stop?

  4. 04

    Cheapest test

    What is the fastest, cheapest real-world test of the riskiest assumption?

  5. 05

    Decision

    Based on the evidence, should you:

    BUILDTEST MOREPARKKILL

What you get

What the founding version gives you

Two things. The system — a reusable workspace you keep and run on every future idea. And one Validation File — one of your ideas taken through that system by hand. These are the sections of that file.

01

Idea Brief + Quick Kill

Clarify the idea and identify early reasons it may not deserve more time.

02

Assumption Map

Identify what must be true and which assumptions are most dangerous.

03

Evidence Ledger

Separate actual evidence from opinions, assumptions and AI-generated claims.

04

Risk / Falsifier Analysis

Define what evidence would seriously challenge the idea before you get emotionally attached to it.

05

Cheapest Test

Choose a practical real-world test for the riskiest assumption.

06

Decision Record

Turn the test result into a written BUILD / TEST MORE / PARK / KILL decision.

Throughout

AI Challenger Prompts

Seven prompts, one per stage, for using a model as an adversary — attacking the idea, auditing your own evidence grades, tightening the falsifier. Each one says what to keep and what to throw away. None of them produce evidence.

And the system itself

The reusable validation workspace — yours to keep

A 62-file Markdown workspace, downloaded at purchase. It opens as an Obsidian vault, or in any text editor. The method itself, so you can run it again without me on every idea after this one.

  • 13 foldersOne per stage, from defining the idea to the decision
  • 9 templatesBlank, copyable versions of every artifact
  • 5 worked examplesOne per decision, plus one full Validation File
  • 7-day sprintA structured week, if you want a schedule

No plugins, no accounts, no subscription. Several of the worked examples end in KILL.

Sample artifacts

Don't just read the framework. See the work.

Three pages from a Validation File, filled in with example entries. They show the structure of the work — the entries are illustrative, not market benchmarks.

Assumption Map

Example · A-01
Assumption
“Target users experience this problem often enough to seek a solution.”
Evidence
UNKNOWN
Strength
None yet
Risk
High
What would change my mind?
10 qualified conversations with repeated problem language + at least 3 users willing to take a concrete next step.

Test Card

Example · T-01
Riskiest assumption
People will pay for this solution.
Test
Offer the founding version to qualified builders.
Pass condition
At least 3 genuine purchases from qualified strangers.
Kill / change condition
No meaningful willingness to pay after a defined outreach/test cycle.

Decision Record

Example · D-01
Assumption tested
Target users will commit to a paid next step.
Evidence
Problem described unprompted in several interviews. No paid commitment yet.
Strength
Moderate for the problem. None yet for payment.
Decision
TEST MORE
Next test
Offer a paid pilot to the users who raised the problem. Decide again on the replies.

Why this is different

Not another AI-generated idea score.

Generic AI validator

  1. Idea
  2. AI analysis
  3. Score
  4. Report

Validate Before You Build

  1. Idea
  2. Assumptions
  3. Evidence
  4. Falsifier
  5. Test
  6. Decision

An AI can generate a convincing explanation for almost anything.

That doesn't make the explanation evidence.

This system separates

  1. What you think

    Assumptions, beliefs and AI-generated claims.

  2. What you know

    Evidence with a source and a strength.

  3. What you need to test

    The riskiest gap between the two.

AI helps process and challenge the work. It does not get to manufacture proof.

The AI principle

AI can challenge the idea. It can't become the evidence.

Use AI to:

  • find hidden assumptions
  • challenge your reasoning
  • structure messy evidence
  • improve interview questions
  • design cheaper tests
  • challenge your final decision

Don't use AI output as proof of demand.

If a claim cannot be supported by actual evidence, mark it:

UNKNOWN.

This is a core principle of the product.

Who it is for

Built for builders who are about to commit serious time.

Good fit

  • You have a real product idea.
  • You are capable of building it.
  • You are tempted to start coding immediately.
  • You aren't sure whether the evidence is strong enough.
  • You want a decision, not another 40-page market report.
  • You are willing to talk to users, test assumptions and face uncomfortable evidence.

Not the best fit

  • You want someone to tell you your idea is good.
  • You want an AI-generated market report.
  • You want guaranteed validation.
  • You don't want to interact with real potential customers.
Shivam Chaturvedi

About the creator

Built from an evidence-first research discipline.

Shivam Chaturvedi

Strategic Research Consultant

My work is strategic research: competitive intelligence, market intelligence, due diligence, market mapping and AI-assisted research. The discipline is the same every time — verify before asserting, triangulate before trusting, and say plainly what is still unknown.

“I built this from the same evidence-first research discipline I use in strategic research — and I'm applying it to product ideas.”

Validate Before You Build is new. This founding version is its first release, and it will be shaped by the people who use it first.

More about my research work

Founding version

The Founding Version

This is the first release of Validate Before You Build.

The founding version is intentionally small.

You submit your idea and the evidence you already have.

I prepare your initial Validation File manually and deliver it within 48 hours.

Your file focuses on:

  • what must be true
  • what evidence you actually have
  • how strong that evidence is
  • the assumption most likely to kill the idea
  • the cheapest useful test
  • the current evidence-supported decision

Founding Version

$39one-time · one idea

How it works

  1. 1Pay $39
  2. 2Submit your idea
  3. 3I prepare your Validation File
  4. 4Receive it within 48 hours
Get the Founding Version — $39

7-day refund if the delivered file doesn't provide a clear, useful decision about what to do next. The refund covers the usefulness of the file — no outcome for your idea is guaranteed.

How it works

Four steps from idea to file.

  1. 01

    Pay $39

    One-time founding-version purchase. Not a subscription. You get the reusable validation system immediately, as a download.

    Get Started — $39

  2. 02

    Submit your idea

    Tell us what you're building, who it's for, what you've already done and what evidence you have.

    Submit Your Idea

  3. 03

    I prepare your Validation File

    Prepared manually. The idea is structured around assumptions, evidence, risk, falsifiers and the cheapest useful test.

  4. 04

    Receive it within 48 hours

    Delivered manually within 48 hours of your submission.

FAQ

Questions, answered plainly.

Is this an AI idea validator?

No. AI is used to challenge and process the work. AI-generated claims are not treated as evidence.

Do I get a market research report?

No. The goal is a decision about what to test next, not a generic market report.

What happens after I pay?

You download the validation system straight away and can start on your idea immediately. To claim your personalized Validation File you submit your idea and existing evidence through the submission form; it is then prepared manually and delivered within 48 hours of that submission.

Do I get the system itself, or just the file you write?

Both. The system is a 62-file Markdown workspace that opens in Obsidian or any editor — the method, templates, five worked examples and a 7-day sprint. It is yours to keep and reuse on every future idea. The Validation File is one idea of yours run through that method by hand.

What if I never submit an idea?

You keep the system, which is the larger and more reusable half of what you bought. The Validation File only starts when you submit — nothing is triggered by the purchase alone.

Is $39 a subscription?

No. It is a one-time founding-version purchase.

Does this guarantee my idea will succeed?

No. It is designed to help you make a better decision before committing more time and resources.

What if the idea is bad?

That's the point. A useful outcome can be BUILD, TEST MORE, PARK or KILL.

Before you build the next thing, find out what needs to be true.

  • One idea.
  • One riskiest assumption.
  • One cheap test.
  • One clearer decision.
Get the Founding Version — $39

Already purchased? Submit Your Idea