Trust & methodology

Built for evidence, not guesswork.

Pinkk is an intelligence research tool that turns public signals into structured, confidence-rated findings — for creators, brands, and affiliates across any niche. This page covers who built it, why it exists, how reports are produced, how your data and account are protected, and how the business is funded.

Why we built Pinkk

Matt

Founded by Matt

Solo founder, independent, and self-funded. With 15 years as a developer and content creator, we built Pinkk because profile intelligence had a real gap — inflated signals, opaque data, and tools that bury you in dashboards instead of giving you answers.

We started in iGaming and gaming communities — where inflated signals and inauthentic traffic are most aggressive — and built from there into a general-purpose intelligence research tool that works across any niche.

Honesty over confidence

A result that says "we're not sure" is more valuable than one that sounds certain but isn't. We'd rather under-promise and be right.

Reports, not dashboards

Structured, readable reports with confidence-rated sections — not raw data exports that require interpretation.

Versioned over time

Run the same research multiple times and track how a profile changes. Every report is stored and versioned.

Why teams choose Pinkk

Most tools give you a confident-looking score and ask you to take it on faith. We're built the other way around: conservative analysis, confidence-rated findings, and reports anyone can verify publicly.

Pinkk approach Common alternative
Structured intelligence with labelled sections Raw metrics dashboard requiring manual interpretation
Confidence-rated findings per section Single opaque score with no reasoning shown
Publicly verifiable reports with PKK reference numbers Black-box results the recipient must accept on faith
Uncertainty shown explicitly, not papered over Overconfident output regardless of evidence quality
Analyst interprets evidence and decides Automated judgment the analyst is expected to act on

Grades reflect evidence — not verdicts

Pinkk assigns an overall A–F grade based on the strength, clarity, and consistency of evidence. A lower grade may indicate insufficient data or conflicting signals — not that a profile is definitively problematic.

How grading is calculated

The analyst makes the call

Pinkk is not a fraud detector or an identity verification service. It surfaces evidence and makes uncertainty explicit — the analyst weighs the confidence levels and decides. We're designed to amplify professional judgment, not replace it.

See all capabilities

Methodology

Sources we use, how confidence levels are decided, what we will never claim, and why conservatism is a feature.

Data sources — public signals only

Platform metrics

Follower counts, reach, impressions, engagement rates, and content categorisation via platform APIs.

Public web signals

Domain authority, backlink profiles, search presence, and audience geography where available.

Cross-platform comparison

Discrepancies and consistencies between platforms are flagged explicitly rather than averaged away.

Historical records

Prior analysis runs on the same profile are compared to surface material changes over time.

Confidence levels

High

Multiple independent signals corroborate the finding. Public data is complete and current.

Medium

Some signals support the finding but corroboration is partial. Direction is clear; magnitude is uncertain.

Low

Insufficient public data or contradictory signals. Surfaced because it may be relevant, not as a sound basis for a decision alone.

Confidence is calibrated continuously through adjudicated accuracy review and can only be lowered by review feedback, never raised.

Claims we will never make

Conservatism is enforced at report-generation, not just policy: no fraud judgments, no identity verification, no legally binding determinations, no use in decisions that affect individual rights (lending, hiring, housing, insurance), no AI training on subject data, and no resale of report content or profile data.

Why we stay conservative

Every report uses cautious, qualified language throughout: "may indicate", "suggests", "could be." Strong language is reserved for findings that earn it. AI surfaces patterns and structures findings — it does not make the final call. Human reviewers assess findings against raw evidence before they reach a report, and disagreements recalibrate future analyses.

"We'd rather under-promise and be right."

Pinkk founding principle

Security

Sensitive data is encrypted at rest. Reports are access-controlled and never exposed publicly. We do not sell your data or use it to train models.

Encrypted at rest

API keys and credentials use application-level encryption, decrypted only when needed.

Reports stay private

Scoped per account, never served from a public URL. Access requires sign-in or a time-limited signed link.

Strong passwords & breach checks

12+ characters with mixed case, numbers, symbols. New passwords are checked against Have I Been Pwned.

Optional 2FA

TOTP-based two-factor authentication with any standard authenticator app, plus recovery codes.

Disposable email blocking

Registrations from known disposable domains are blocked to keep accounts reachable.

Rate limiting & bot protection

Sensitive endpoints are rate limited. Forms use Cloudflare Turnstile against automated submissions.

Export or delete anytime

Export reports as PDFs and delete your account and data at any time. No lock-in. See our Privacy Policy.

Live system status

99.94% · 30 days

Uptime is monitored externally and published on our public status page.

Responsible disclosure

Found a security issue? Report it privately through our contact page before public disclosure, and give us a reasonable window to investigate and fix. We will work with you in good faith.

Transparency

Pinkk is an early-stage product. This section covers how we are funded, which infrastructure we rely on, and how we use AI. We keep it current as things change.

Funding

Primarily funded by its founder — personal capital, no outside investors, no venture funding. We also use startup infrastructure credits from AWS, Cloudflare, and Notion; these are not investment, and none of these partners hold equity, influence the roadmap, or have editorial say in how reports are written.

How we use AI

Which models

Reports are processed using models from OpenAI and Anthropic via their commercial APIs. Sable — our in-house intelligence layer — coordinates provider routing, governance, and confidence controls rather than coupling the platform to one vendor.

Your data & model training

Data sent to OpenAI and Anthropic is not used to train their models — a contractual protection in their API terms. We do not share account data or report contents beyond what's necessary for a given analysis.

Limitations

AI can be wrong. Models can hallucinate or reflect training bias. Every finding carries a confidence rating — uncertain findings are flagged, not dressed up as facts.

No automated decisions

Reports inform human decisions — they don't trigger automated actions. A report never blocks an account or takes action in a third-party system on your behalf.

Infrastructure & dependencies

Hosting

AWS infrastructure, databases managed and backed up daily.

CDN & security

Cloudflare in front of all traffic for DDoS protection, caching, and bot filtering.

Payments

Billing via Polar.sh. We never store card details.

Email & monitoring

Transactional email via Resend. Uptime via BetterStack, visible on our status page.

Last updated

This page was last reviewed in September 2026. We update it whenever the material facts change. Questions or corrections? Contact us.

Try it for free.

Create a free account, add a profile, and run your first intelligence report. No credit card required.