Products
Everything on the platform, and how far along it is.
13 available today, 24 more in testing, beta, or build. Each one is labelled with its real status.
Available13
Shipped. You can use these today.
SynOI Decisions
Cryptographically-signed audit trail for every gateway-routed AI action. Publicly verifiable.
Learn more →AvailableReceipt Verifier
No-auth verification against the published public keys, or offline from an exported bundle. A full check needs RFC 8785 JCS canonicalization plus both Ed25519 and ML-DSA-65 verifiers, not Ed25519 alone. The @synoi/verify CLI receipt-ID mode fetches from the gateway (internally consistent, not offline). Hosted paste-a-URL page at oid.synoi.systems is planned, not yet deployed.
Learn more →AvailablePour
Pour checks that a copy of your data matches its source, row by row and value by value, and names exactly what doesn’t. Your data stays on your own infrastructure by default. Reads JSON, CSV, Parquet, Excel, Postgres, Oracle, SQL Server, MySQL, MariaDB and SQLite, and the same files in S3, Azure Blob or Google Cloud Storage; declare a key and row order stops mattering, so a migration can be checked table by table with row counts and control totals. It compares the data itself, so the same table in JSON, CSV, Parquet, Excel or a database still matches.
Learn more →AvailableSynOI Gateway
Govern any action you route through one gateway. One call, or one env var for LLM traffic.
Learn more →AvailableSynOI Vault
The content-addressed store behind every SynOI product. Two tiers in one substrate: a working memory tier (HNSW-indexed semantic match, LRU eviction) and a canonical tier (authoritative, append-only, point-in-time queryable). Backs the gateway cache, stores Decision Receipts, powers RAG retrieval, content-addressed and signed end to end.
Learn more →SynOI Conductor
Capability registry and approval routing: what your AI can call, who approves it, and when. Ships inside the grant engine; the standalone Conductor daemon is retired.
@synoi/gap
The open, vendor-neutral format that lets any agent prove what it was allowed to do. Type library for the Governed Action Protocol: declarations, grants, invocations, receipts.
Learn more →Available@synoi/sraid
The low-level building block every SynOI receipt is made of; most teams never touch it directly. SRAID L0 object layer: CDRO construction, OID derivation, canonical JSON, hybrid signing.
Learn more →AvailableSynOI MCP Server
A bridge that lets any AI assistant use your governed tools and knowledge. Model Context Protocol (MCP) server exposing your governed Persona, Skills, and knowledge base to Claude Desktop, Cursor, or the MCP CLI.
Learn more →AvailableFor MSPs
Multi-tenant governance for managed service providers.
Learn more →AvailableSupply Chain Guard
Catch the self-propagating package worm before it poisons your AI toolchain. State-aware detection across versions.
Learn more →SynOI OID Resolver
Published open-source library (@synoi/oid-resolver) for content-addressed package attestation and OID resolution. The hosted, DNS-like lookup service is still roadmap; the library ships today.
Post-Quantum Migration
Hybrid Ed25519 + ML-DSA-65 signing, shipped in the Node/CLI receipt verifier; the hybrid signature is checkable offline against a key you hold (exported bundle or your own library). Browser and extension builds are Ed25519-only for now.
Testing11
Built and running in our own testing. Not yet open to everyone.
Multi-Surface HITL
Human approval on dangerous tool calls via desktop, Slack, SMS, or browser.
Risk Policy Engine
Per-tenant allow / deny / require-approval rules. Evaluated in <1ms.
MCP Gateway
Govern Model Context Protocol tool calls transparently.
Routing Profiles
Named routing presets you configure per environment (auto, eco, premium, free, reasoning). Automatic cost-optimized selection between them is not built yet.
@synoi/sdk
One-line HITL governance for any AI agent. Vendor-agnostic.
Learn more →TestingSynOI for OpenClaw
Native OpenClaw plugins (router + guard). LLM keys never leave your machine.
Learn more →TestingSynOI Connect
One command points Claude Code, Cursor, Aider, and OpenClaw at your SynOI gateway. Zero dependencies, routing only: the hosted brain decides, your local tools stay in your custody.
Learn more →Testing@synoi/langchain
Governance interceptor for LangChain, AutoGen, and CrewAI agents: every tool call produces a Decision Receipt through your gateway before it runs.
Learn more →Desktop Helper
Native OS approval notifications: Windows, macOS, Linux.
SynOI Quickstart
docker compose up → working gateway in 5 minutes.
Slack HITL Channel
Approve dangerous AI actions from the same thread Viktor lives in.
Beta4
Functional and still changing.
Policy Packs
Pre-built rule sets for HIPAA, SOC 2, PCI, custom risk policies.
CI/CD Governance
Sign every Terraform apply, every deploy, every release.
SynOI Edge
On-prem governance. Your AI traffic never leaves your network.
SynBox
Full-capability edge node. Docker container. Offline-capable.
Coming Soon9
In build, with a target window.
Batch-Seal
One signature sealing many receipts at once: evidence for a whole batch of decisions in a single anchor.
SynOI Cloud
A SynOI-hosted gateway mode, one env var instead of running your own. Not live yet: production deployments today are self-hosted (Edge).
SynOI Token Saver
Cuts redundant tokens on gateway-routed traffic; early access.
Learn more →SynAssist
Lightweight edge sidecar. Policy enforcement at the agent layer.
@synoi/guard
Pre-execution governance hook for Claude Code and Cursor. Pre-release.
Learn more →SynOI Mobile
React Native shell + module store. Governance on your phone.
Mobile HITL Module
Biometric approve / deny push notifications from anywhere.
On-Device AI
Chat, transcription, translation, OCR: all on your NPU, all private.
SynOI Exposure Scan
Free public scan: discover exposed Ollama, n8n, Flowise, and AI proxy endpoints on your network. Lead-gen + public-good infrastructure for the agentic security category.
18 more are specced and not yet in build. They are not listed here, because a list of things we have not started is not a list of products.
New here? Start with the ten core capabilities.