Switchboard Product Overview

What Switchboard Is

Switchboard is a multi-tenant MCP (Model Context Protocol) gateway that gives AI assistants secure, audited access to business tools through a single connection point.

Instead of AI tools connecting directly to Google, Slack, QuickBooks, etc., they connect through Switchboard — which handles authentication, logging, and security in one place.

The Problem We Solve

For Developers / Power Users

Setting up MCP servers is painful — each service needs its own OAuth app, credentials, and config
Claude Code's built-in tools are limited and don't support all the services teams actually use
No centralized way to manage which AI tools can access what

For IT / Security Teams

Shadow AI is already happening — developers are using Claude Code, Cursor, and custom MCP clients
Zero visibility into what data AI is accessing
No audit trail for compliance (SOC 2, HIPAA, GDPR)
Enterprise solutions cost $100K+/year and require complex deployments

For Organizations

Want AI productivity gains but can't ignore governance
Need something between "no controls" and "full Zero Trust deployment"
Can't afford 6-month implementation projects

What's Built Today

Core Platform

Feature	Description	Status
Multi-Service OAuth	Single sign-on to 27 services via Switchboard	✅ Production
~992 MCP Tools	Comprehensive tool coverage across all connected services	✅ Production
Multi-Account Support	Connect up to 5 accounts per service (work + personal)	✅ Production
Unified Token	One bearer token for all services	✅ Production
Per-Service Tokens	Granular tokens for specific services	✅ Production
Dashboard UI	Web interface for connection management	✅ Production

Connected Services (27 Total)

Productivity & Communication

Google Workspace
Slack
Notion
Teamwork

Sales & CRM

HubSpot
Apollo.io
PandaDoc

Finance & Accounting

QuickBooks

Infrastructure & DevOps

Cloudflare
GitHub

Marketing & Analytics

Meta (FB, Instagram)
Fathom Analytics
ManyChat

Other

WordPress
n8n
Figma

Security & Compliance (Already Built)

Feature	Description	Status
Full Audit Logging	Every tool call logged with user, service, tool, timestamp, latency	✅ Production
Content Classification	Automatic detection of PII, PHI, financial data, credentials	✅ Production
Credential Redaction	API keys/tokens automatically redacted before storage	✅ Production
R2 Content Storage	Full request/response bodies stored for compliance	✅ Production
Multi-Tenancy	Per-organization isolation with role-based access	✅ Production
Legal Hold	Prevent deletion of audit data during investigations	✅ Production
Configurable Retention	Per-tenant retention policies (30/90/365 days)	✅ Production
Alert Rules	Anomaly detection and threshold-based alerts	✅ Production
Admin Panel	Compliance dashboard, user management, audit viewer	✅ Production

Classification Patterns (What We Detect)

Category	Detection Method	Examples
Credentials	Regex patterns	API keys, passwords, tokens, private keys, AWS keys, GitHub tokens
PII	Regex patterns	SSN, email, phone, credit cards, date of birth
PHI	Keyword matching	Patient data, diagnoses, prescriptions, medical records
Financial	Regex patterns	Account numbers, routing numbers, tax IDs

Technical Architecture

100% Cloudflare Stack

Workers — Compute, MCP handling, OAuth

D1 — User data, credentials, audit logs

R2 — Request/response content storage

Queues — Async audit processing

Access — Dashboard authentication

Why This Architecture Matters: No Kubernetes Required

Every competitor in this space requires Kubernetes infrastructure. Switchboard is the only solution that doesn't—because Cloudflare Workers eliminates the need for container orchestration entirely.

What Competitors Require	Switchboard Alternative
Kubernetes cluster ($100-500/mo)	Serverless—no infrastructure cost
DevOps team ($150K+/year)	No DevOps required
2-6 week deployment	Deploy in hours
Capacity planning & node sizing	Automatic infinite scaling
Ongoing K8s maintenance	Zero maintenance—fully managed

Result: Mid-market companies (200-1000 employees) can get enterprise AI governance without enterprise infrastructure costs or complexity.

Security

AES-256-GCM encryption for all stored tokens
PBKDF2 key derivation with per-deployment salt
Bearer token authentication for API access
Cloudflare Access for dashboard protection

Use Cases

Use Case 1

The Solo Developer

Persona: Freelance developer using Claude Code for client projects

Problem: Needs to access client's Google Drive, Slack, and QuickBooks from Claude Code, but each client has different accounts, doesn't want to store client credentials locally, needs to switch between projects seamlessly.

Solution: Connect all client accounts with aliases. Single Switchboard token in Claude Code config. Specify which account in tool calls: account: "client-a". Audit trail shows which client data was accessed when.

Value: Hours saved on setup, professional credential management, clear audit trail for billing disputes.

Use Case 2

The Growing Startup (20-50 employees)

Persona: Head of Engineering at a Series A startup

Problem: Developers using Claude Code and Cursor with no visibility. SOC 2 auditor asking questions about AI governance. Can't afford enterprise security tools.

Solution: All developers connect through Switchboard. Central admin sees all tool usage. Compliance reports show what data AI accessed. Per-developer access controls.

Value: SOC 2 compliance checkbox, visibility without slowing down developers, 1/10th the cost of enterprise alternatives.

Use Case 3

The Mid-Market IT Team (200-500 employees)

Persona: IT Director at a manufacturing company

Problem: Shadow AI is everywhere. CEO read an article about AI data leaks and is asking questions. Need visibility without blocking productivity.

Solution: Deploy Switchboard as the "official" way to use AI tools. Route all AI agent access through the gateway. Instant visibility. Policy controls. Audit trail satisfies executive concerns.

Value: Governance without disruption, executive-ready reports, foundation for future Zero Trust deployment.

Use Case 4

The Compliance-First Organization (Regulated Industry)

Persona: CISO at a healthcare or financial services company

Problem: Strict compliance requirements (HIPAA, SOC 2, PCI). AI tools are being used but governance is unclear. Need to prove AI isn't accessing PHI/PII inappropriately.

Solution: Content classification flags PHI/PII in real-time. Legal hold prevents data deletion during audits. Detailed audit logs with full request/response content. Compliance reports exportable for auditors.

Value: Audit-ready AI governance, defensible compliance posture, risk mitigation.

Future Roadmap

Phase 1 Security Hardening (Weeks 1-4)

Pre-execution DLP — Block sensitive data BEFORE tool execution
Policy Rules Engine — Configurable rules per org (block/warn/allow)
Custom Block Responses — User-friendly error messages
Source Code Detection — Add code patterns to classification

Phase 2 Cloudflare Integration (Weeks 5-8)

AI Gateway Integration — Route AI inference through CF AI Gateway for DLP
Tenant API Integration — Create per-org Cloudflare accounts automatically
CF Analytics Bridge — Per-org analytics dashboards
DNS Gateway (Optional) — Enable CF Gateway DNS filtering for orgs

Phase 3 Platform Maturity (Weeks 9-12)

MCP Server Remixing — Let users combine/customize tool sets
Comprehensive Testing — Full test coverage, E2E tests, load testing
Codebase Cleanup — Refactoring, documentation, code review
SIEM Export — Push logs to customer's SIEM/SOAR

Phase 4 Go-to-Market (Weeks 13-16)

Pricing & Packaging — Finalize tiers (Free/Pro/Enterprise)
Marketing Site — Public-facing product page
Case Study Documentation — 3-4 detailed use cases with metrics
CF Partner Pitch — Materials for Cloudflare partnership discussion

Competitive Landscape

The Key Finding: Nobody is Targeting Mid-Market

We analyzed four emerging MCP gateway/security competitors. None of them serve the mid-market without requiring Kubernetes infrastructure.

Competitor	Funding	Target	Deployment	Mid-Market?
Microsoft MCP Gateway	Microsoft-backed	Azure enterprise	Kubernetes	❌ No
Obot	$35M (Rancher founders)	Enterprise/DevOps	Kubernetes	❌ No
Agentic Community	Open source	AWS enterprise	Kubernetes	❌ No
TrueFoundry	$21M	Enterprise ($100K+)	On-prem/K8s	❌ No

Our Unique Position

No Kubernetes required — Serverless on Cloudflare Workers
Deploy in hours, not weeks — No infrastructure to manage
Mid-market pricing — $500-2K/mo, not $100K+/yr
Compliance-ready — SOC 2, HIPAA, GDPR out of the box
Cloudflare ecosystem — Natural upgrade path to full Zero Trust

See full competitive analysis →

Success Metrics

Metric	Target	Why
Connected Organizations	10 in 90 days	Adoption
Active Users	50 in 90 days	Engagement
Tool Calls/Day	1,000+	Usage depth
Blocked Requests	Track but no target	DLP effectiveness
Conversion to Managed Security	20% of paid users	Lead gen validation
CF Expansion Revenue	$20K+ ARR	Partnership value

Technical Reference

API Endpoints

switchboard.fpdigital.ai (Dashboard - CF Access protected)
├── /dashboard                    → Main UI
├── /dashboard/services/:svc      → Service management
├── /admin                        → Admin panel
└── /guides/:service              → Setup guides

api.switchboard.fpdigital.ai (MCP API - Bearer token auth)
├── /mcp/unified                  → All tools (unified token)
├── /mcp/{service}                → Per-service tools
└── /oauth/*/callback             → OAuth callbacks

Token Formats

Token Type	Format	Use Case
Unified	`fps_unified_xxx`	All services, one token
Per-Service	`fps_google_xxx`	Single service, all accounts
Per-Account	`fps_google_work_xxx`	Single service, single account