Skip to content
Graph of Plan

Nomos Cloud

Nomos Cloud is the hosted platform for domain-driven API generation and agent execution auditing. It makes AI systems auditable and debuggable via append-only decision traces.

The Short Version

Every AI agent takes actions. Most leave no trace of why. Nomos Cloud is an agent execution gateway that emits “chains of truth”—append-only, queryable decision traces that capture not just what happened, but why it was allowed. Think of it as event sourcing for AI: every intent, directive, policy evaluation, and approval becomes a permanent, auditable record.

The Problem

The industry is waking up to context graphs—the idea that the most valuable thing you can store isn’t the AI’s chain-of-thought, but the decision context: who requested it, what policy approved it, what precedent justified it, and what artifacts it produced.

Most agent systems either:

  • Have no audit trail at all
  • Log raw text that’s impossible to query
  • Lose context when sessions end

Nomos captures the decision graph as first-class data, making every agent action traceable back to its authorization and forward to its consequences.

Our Approach

Event-Sourced Architecture

Nomos is built on event sourcing principles: every change is captured as an immutable event. This means:

  • Deterministic replay: Rebuild any state by replaying events
  • Local-first: Work offline, sync when connected
  • Full auditability: Every decision is traceable

Core Concepts: Intents → Directives → Events

The data model follows a clear flow:

  1. Intents: What the user or agent wants to do
  2. Directives: What actions are permitted based on policy
  3. Events: What actually happened, with full context

This isn’t a blockchain—there’s no global consensus. Instead, Nomos models many overlapping, queryable chains by workspace, run, timeline, correlation, or entity.

Stories, Not Logs

A normal audit log is a crime against human attention. Nomos stores templates and variables, rendering human-readable stories on the read path:

“Sarah asked: Who owes us money? Agent queried Xero. Found 12 overdue invoices totaling £47,230. Agent drafted reminders. Finance approved. Messages sent.”

The raw structured trace remains canonical—stories are just a view. This avoids persisting PII-heavy narratives while keeping the audit trail complete.

How It Works

OpenAPI/MCP Generation from Domain Models

Define business domains in a typed, validated format. Nomos generates:

  • REST APIs with OpenAPI specs
  • SDKs in multiple languages
  • CLI tools
  • MCP tools for AI agent integration

Everything compiles from the domain model, eliminating drift between documentation, API, and implementation.

Cloudflare-Native Hosting

Built entirely on Cloudflare primitives:

  • Workers: HTTP APIs
  • Durable Objects: Per-timeline sequencing
  • R2: Immutable ledger storage
  • Queues: Async indexing
  • Workflows: Heavy derivations
  • D1: Query indexes

Every tenant gets logically isolated storage. Compute scales to zero. No Kubernetes to manage.

Adapters for Everything

The SDK includes adapters that transform existing telemetry into Nomos entries:

  • OpenTelemetry spans become events with correlation preserved
  • OpenAI and Anthropic tool calls become directives
  • MCP tool invocations become policy-checked directives

You don’t rewrite your agent—you instrument it with a few lines.

Product Features

@nomos/router-sdk

The open-source TypeScript SDK: entry schema definitions, ID generation (UUIDv7 for time-sortable IDs), story template helpers, safe redaction utilities, and adapters. MIT-licensed.

Nomos Cloud Service

The managed service: ingest API, query API, story rendering, web dashboard, and all compliance/enterprise features. Customers get a workspace URL, drop in the SDK, and start capturing traces immediately.

Web Dashboard

Timeline visualization showing entries as a scrollable, zoomable graph. Story view rendering human-readable narratives. Chain explorer for following correlation links. Entity history showing all actions affecting a specific customer/account/resource.

Who It’s For

AI Agent Developers

Anyone building autonomous systems needs to understand what their agents did and why, debug unexpected behavior, and demonstrate compliance. Nomos gives them a single place to capture and query decision traces across their entire agent fleet.

Regulated Industries

Finance, healthcare, legal, and government need audit trails that survive legal scrutiny. Nomos provides tamper-evident logs, cryptographic proofs, and the ability to reconstruct exactly what happened at any point.

Enterprise AI Teams

Large organizations deploying AI across multiple teams need centralized visibility: which agents are running, what actions they’re taking, which policies are being applied. Nomos is the control plane for enterprise AI governance.

Business Model

The Neo4j Model

The underlying technology—the ledger schema, the SDK, the adapters—is fully open-source under a permissive license. Anyone can run Nomos on their own infrastructure.

The business model is convenience: businesses pay for hosted infrastructure, compliance guarantees, a web dashboard with visualizations, and enterprise features they don’t want to build themselves.

Pricing

  • Free Tier: Generous limits for hobbyists and startups
  • Usage-Based: Pay for entries stored and queries executed
  • Enterprise: SOC 2, SSO, RBAC, custom retention, dedicated support

Pricing Rationale

Competitor benchmark:

ProductPriceModel
LangSmith£31-400/moUsage-based tiers
HeliconeFree-£12-120/moUsage-based
Datadog APM£25+/host/moPer-host + usage
New RelicUsage-basedPer-ingest volume

Our positioning:

Nomos Cloud charges for value delivered: entries stored and queries executed. This scales naturally with agent deployment—small experiments are cheap; production workloads pay proportionally.

Value anchor:

  • ARPU: £150/mo for production usage
  • Compare to: Building internal audit infrastructure (£50k+ engineering time)
  • Compare to: Compliance audit failure (£100k+ fines, reputation damage)
  • Breakeven: One avoided compliance incident pays for years of Nomos

Price confidence:

  • Free tier enables SDK adoption without friction
  • Usage-based grows with customer value
  • Enterprise tier captures compliance requirements
  • 90% gross margin supports sales investment

Compliance & Certifications

For regulated industries: SOC 2 Type II certification, GDPR data processing agreements, HIPAA BAAs for healthcare, and custom data residency options. Tamper-evident hash chaining with cryptographic proofs available for audit.

Unit Economics

Revenue Formula

Monthly Revenue = (SDK Conversions × ARPU) + Enterprise MRR


SDK conversions:
  Downloads/mo × Conversion rate = Customers
  1,000 × 3% = 30 new customers/mo


Revenue mix:
  Usage-based:  Customers × £150/mo ARPU
  Enterprise:   1 deal/quarter × £25k ACV = £2,083/mo

Cost Structure

TypeAmountNotes
Base ops£5,000/moPlatform, support, compliance
Per-customer£15/moCloudflare compute, R2 storage

Unit margin: £150 ARPU - £15 infra = £135 gross profit (90% margin).

Key Metrics

┌─────────────────────────────────────────────────────────┐
│  ARPU: £150       │  Margin: 90%      │  LTV: £4,500   │
│  CAC: £100        │  Payback: <1mo    │  LTV:CAC: 45:1 │
│  Conversion: 3%   │  Enterprise: £25k ACV              │
└─────────────────────────────────────────────────────────┘

Acquisition Strategy

Primary channel: Open-source SDK → hosted conversion (the Neo4j model)

The £100 CAC reflects a developer-led, open-source-first acquisition model:

Channel% of AcquisitionWhy It Works
Open-source SDK adoption50%MIT-licensed SDK gets usage; cloud converts convenience buyers
Agent framework integrations25%Adapters for LangChain, AutoGPT, CrewAI bring SDK to existing users
Technical content15%“Event sourcing for AI agents” positions Nomos as the authority
Enterprise inbound10%Compliance teams searching for “AI audit trails” find Nomos

The Neo4j playbook:

  1. Open source first: SDK is MIT-licensed, anyone can self-host
  2. Cloud convenience: Managed service removes operational burden
  3. Enterprise pull: Compliance requirements make hosted version necessary
  4. Land and expand: Start with one team’s audit needs, expand across the org

Why this CAC is achievable:

  1. Zero-friction trial: SDK download → cloud sync is one config change
  2. Developer advocacy: Every SDK user is a potential advocate to their org
  3. Enterprise deal size: £25k ACV makes higher-touch sales economical
  4. Regulatory tailwinds: AI governance requirements create demand

Conversion funnel:

  • SDK downloads: 1,000/month (target)
  • Free tier signup: 30% of downloaders
  • Paid conversion: 10% of free tier
  • Enterprise upgrade: 1 deal/quarter at £25k ACV

Proof points needed: 3 enterprise logos using Nomos for compliance audit trails.

Year 1 Projection

MonthCustomersUsage MRREnt MRRTotalExpensesNetCumulative
M130£4,500£0£4,500£5,450-£950-£950
M390£13,500£2,083£15,583£6,350£9,233£11,400
M6180£27,000£4,166£31,166£7,700£23,466£65,500
M12360£54,000£8,333£62,333£10,400£51,933£302,000

Assumes: 1,000 SDK downloads/mo, 3% conversion, 1 enterprise deal/quarter.

Sensitivity

ScenarioM12 CustomersM12 MRRY1 Net
Base (3% conv, 1 ent/qtr)360£62k£302k
Higher conversion (5%)600£98k£520k
No enterprise360£54k£230k
Slower downloads (500/mo)180£35k£140k

Roadmap

M0: Schema & Conventions

Define the canonical ledger entry schema, ID conventions, and story template format. Publish the TypeScript SDK.

M1: Ingest & Persistence

Implement the ingest API, per-timeline sequencing via Durable Objects, and append-only storage in R2.

M2: Query & Story Rendering

Add the query API with filtering by runId, correlationId, entityRef, and time range. Implement story rendering.

M3: Dashboard MVP

Build the web dashboard with timeline visualization, story view, and chain exploration.

M4: Multi-Tenancy & Billing

Implement workspace provisioning, usage metering, Stripe billing, and free tier.

M5: Compliance & Enterprise

Add tamper-evident hash chaining, SOC 2 controls, SSO/RBAC, and audit-of-audit logging. Pursue SOC 2 Type II certification.

Year 2: Growth Phase

Product expansion:

  • Advanced query language for complex trace analysis
  • Anomaly detection (automatic flagging of unusual agent behavior)
  • Policy engine (define rules, auto-enforce on new entries)

Target metrics:

  • 500+ paying customers
  • £100K MRR
  • 5 enterprise contracts
  • 2 compliance certifications (SOC 2 + ISO 27001)

Year 3: Scale Phase

Platform maturity:

  • Multi-region data residency (EU, US, APAC)
  • Real-time streaming API for live dashboards
  • Regulatory reporting templates (EU AI Act, industry-specific)
  • Self-hosted enterprise edition

Revenue targets:

  • £500K MRR (highest revenue product due to enterprise ACV)
  • 1,000+ SDK installations in production
  • 20+ enterprise contracts
  • HIPAA certification for healthcare vertical

Enabled By

Uses Tooling

Target Customers

Competes With

Underpinning Assumptions