Yan Malinovskiy

Relpin

Internal business apps at AI speed — governed, isolated, promotable.

A platform for building and running internal business apps with a governed, code-first AI builder. Not low-code: humans and AI agents author real TanStack Start / TypeScript codebases that Relpin previews, builds, and deploys on managed infrastructure — with every edit flowing through the same authorization and audit path.

• Database-per-organization isolation with per-environment schemas
• Per-tenant deployments on Cloudflare Workers for Platforms
• Content-addressed releases with gated DEV → TEST → PROD promotion
• Evidence Fabric: hash-chained, verifiable audit across every mutation

Veritio

Tamper-evident audit infrastructure as a service.

A hosted evidence and audit-trail platform: managed ingest with scoped API keys, a provenance graph over every recorded event, and a hosted MCP endpoint so AI agents can query evidence directly. Each tenant gets an isolated Postgres database, provisioned automatically and served from the edge.

• Database-per-tenant on Neon with automated provisioning
• Edge-native on Cloudflare Workers with Hyperdrive and Queues
• Hosted MCP endpoint for AI-agent evidence querying
• Deterministic risk scoring and retention as scheduled jobs

Veritio SDK

A protocol-first evidence layer for application audit trails.

The open-source core of Veritio: a language-neutral spec for hash-chained, tamper-evident audit records, implemented with full protocol parity in TypeScript, Python, and Go — verified by shared conformance fixtures. Ships with framework adapters, host-injected storage for five databases, and a local Workbench with an MCP server.

• Cross-language parity (TS / Python / Go) enforced by conformance fixtures
• SHA-256 hash-chained, append-only records with tenant-scoped idempotency
• Ten-plus framework adapters, from Better Auth to Claude Code
• Local Workbench + MCP server — inspect evidence without a hosted account