Episode 1 — Rebuild the Stack: An API-First Disposable Email Verifier

Every day, thousands of fake users sign up for your product.

They use mailinator.com. guerrillamail.com. tempmail.io. They abuse your free trial, inflate your metrics, and poison your CRM — and your signup form lets them walk right in.

This is not a niche problem. It's a silent tax on every product with an email gate.

Why Existing Solutions Are Failing

Most "email verification" tools do one of two things:

1. Check that the @ symbol exists and the domain resolves
2. Maintain a static blocklist of known disposable providers

The first is trivially useless. The second sounds reasonable — until you realise that disposable email services spin up new domains faster than any static list can track. It's a cat-and-mouse game, and blocklists are losing.

"The problem with fighting disposable emails using a list is that the list is always yesterday's intelligence."

The services abusing your product aren't using the same ten domains from 2019. They're generating fresh ones constantly — and walking straight through every blocklist on the market.

A Different Approach

For Episode 1 of Rebuild the Stack, I'm building an AI-powered Disposable Email Verifier — not as a SaaS dashboard, not as a subscription with seat limits, but as a pure API endpoint.

You send an email address. You get a verdict.

Here's what makes the approach different:

Instead of asking "is this domain on a list?", the model asks "does this domain behave like a disposable service?" — a fundamentally harder question to game.

What Gets Built

The architecture is intentionally lean. No dashboard. No onboarding flow. Three things:

1. A REST endpoint — POST /verify accepts an email, returns a structured verdict with confidence score and detection signals
2. A caching layer — domain-level results cached in Redis so repeat lookups on Gmail, Outlook, and other high-traffic domains return instantly
3. Clean documentation — predictable schema, explicit error codes, no surprises buried in a pricing page

{
	"email": "user@disposable-domain.xyz",
	"verdict": "disposable",
	"confidence": 0.97,
	"signals": {
		"domain_age_days": 3,
		"mx_catchall": true,
		"pattern_match": true
	},
	"cached": false,
	"response_time_ms": 84
}

Sub-second responses, designed to sit inline at your signup flow without adding friction.

Built for Developers and AI Agents

Most verification tools assume a human is operating them through a UI. This one doesn't.

If you're building AI workflows that ingest leads, verify contacts, or clean CRM data — the endpoint plugs straight in. Predictable inputs, structured outputs, explicit confidence scores your agent can branch on. No session management. No dashboard state. Just an API call.

What I'll Share Publicly

This build is happening in the open. Every episode of Rebuild the Stack includes the full decision trail — not just the finished product:

• Why I structured the confidence score the way I did
• What I considered building and deliberately cut
• Where the AI model struggles and how I'm handling edge cases
• The tradeoffs between latency and detection accuracy

Software is full of people sharing polished outcomes. The decisions made along the way are more useful.

Launch Is Coming Soon

This is the tool I wished existed every time I built a product with an email gate. So I'm building it.

If you've ever been burned by trial abuse, fake signups, or a CRM full of @mailinator.com addresses — I'd genuinely like to hear how you handled it. Drop it in the comments.

Follow along to catch the launch when it drops.