Skip to main content
About
SONATE

Building the evidence layer
for operational AI.

I'm Stephen Aitken, founder of YSEEKU and builder of SONATE. I spent twenty years in regulated fintech operations, where audit trails, accountability, and controls were not abstract concepts — they were how the business survived scrutiny. SONATE is the infrastructure I believe AI now needs.

What I'm building

SONATE is a cryptographic audit layer for AI systems.

As AI moves from experiments into regulated workflows, organisations need a way to prove what happened after the fact — not just trust a dashboard, transcript, or vendor-controlled log.

What the system was asked
What it returned
Which policy evaluated it
Whether it complied, escalated, or required review
Who can verify the record independently

SONATE generates cryptographically signed Trust Receipts for every interaction — independently verifiable by regulators, clients, and counterparties.

The Operator

Stephen Aitken spent two decades in regulated fintech operations — managing teams, navigating audits, building processes that had to survive regulatory scrutiny.

That background shaped a specific instinct: if something consequential happens, you need a verifiable record. Not a log. Not a screenshot. Evidence.

When AI started making real decisions in production systems — approving loans, drafting clinical notes, generating legal analysis — the absence of execution evidence became obvious to anyone who'd ever sat across from a regulator.

That is the founder-market fit: operations, risk, compliance, and decision systems were the environment I came from before AI made the evidence problem urgent.

The Insight

We have TLS for networks. Code signing for software. Digital signatures for financial transactions. But AI had nothing — no standard way to prove what a model received, what it returned, or whether it complied with policy.

Non-repudiation is a solved problem everywhere except AI execution.

SONATE exists to close that gap: cryptographically signed, hash-chained Trust Receipts that anyone can verify independently, without trusting the platform that generated them.

The shift from experimental AI to operational AI creates a new requirement: verifiable auditability.

Built solo. Powered by AI.

SONATE was built entirely through AI-assisted development — a solo founder with no traditional engineering background, using LLMs as the development team.

The result: 200,000+ lines of production code across a 20-package TypeScript monorepo, Python SDKs, a 30-page enterprise dashboard, and live cryptographic infrastructure — shipped in under twelve months.

This isn't a limitation. It's proof of the thesis.

If one person can build production infrastructure this complex using AI, imagine what autonomous AI systems are doing inside enterprises right now — without any verifiable record.

20
Core packages
28+
Backend services
30+
Dashboard pages
3
SDKs (TypeScript, Python, Browser)

Where we are

The platform is built. The cryptographic infrastructure is live. Trust Receipts are being generated and verified. The open-source SDKs are published.

The next step is intentional distribution: design partners in regulated industries, investors who understand AI infrastructure, and operators who need evidence when AI systems make consequential decisions.

We're raising pre-seed capital to fund the first enterprise design partners, hire engineering and go-to-market, and move from working product to operating company.

Melbourne, Australia · Founded 2025

What's live today

Ed25519 cryptographic signing + SHA-256 hash chains
W3C DID identity for agents and platform
MIT-licensed verification SDKs (TS, Python, browser)
Deterministic default policy evaluator (replaceable, bring-your-own supported)
Behavioral drift and manipulation signals as advisory evidence
Reference enforcement loop with advisory and explicit-authority modes
Multi-model gateway (OpenAI, Anthropic, Gemini)
Enterprise dashboard with 30+ pages
Real-time compliance reporting and alerting
Interactive demos and browser receipt verifier

Let's talk

I'm currently speaking with investors interested in AI infrastructure, design partners in regulated industries, and operators responsible for AI governance.