The governance layer for enterprise AI

One gateway to govern every AI request across every provider.

Sunesix is the multi-tenant AI Gateway that lets organizations connect OpenAI, Anthropic, Gemini, Azure OpenAI, DeepSeek, Grok, and Ollama behind one governed funnel — with per-employee token limits, full audit trails, and real cost tracking on every request.

Book Demo

No credit card required · Already have a workspace? Sign in

Loading signups…

Lower AI spend
38%

Lower AI spend

Requests audited
100%

Requests audited

Providers, one gateway
7

Providers, one gateway

app.insights.dev/company/ai-gateway

Live request flow

Streaming

Employee prompt

Human-set

Policy check

Provider routed

Response logged

chat-panel.tsx sends the request

GPT-4o

Model routed

142ms

Latency

$0.0021

Request cost

Works with every major AI provider

OpenAI
Anthropic
Gemini
Azure OpenAI
DeepSeek
Grok
Ollama

The AI era has arrived. But organizations need more than another AI tool — they need control of the ones they already use.

The Platform

An AI governance platform, not just an API proxy.

Sunesix lets organizations safely manage every AI provider through one gateway — with the multi-tenant, security, and billing primitives enterprise rollouts actually need.

01

Multi-Tenant SaaS

Every company gets a structurally isolated workspace — enforced at the query layer, not by convention.

02

AI Gateway

One funnel for every provider. Business code never touches a provider SDK directly.

03

Enterprise Security

Encrypted credentials at rest, dual-portal auth, zero-trust tenant boundaries.

04

RBAC

Roles and permissions are data, not hardcoded checks — define custom roles per company.

05

Module Marketplace

Install, enable, and configure business modules with zero downtime and zero redeploys.

06

Cost Optimization

Real, provider-reported token cost logged on every request — not estimated.

07

Audit Logs

Every sensitive action — logins, permission changes, AI requests, billing — is logged and reviewable.

08

Analytics

Live usage dashboards down to the individual employee and model.

09

Billing

Plan-based and module-based billing, with a configurable bundled token allowance.

10

Team Collaboration

Department structures, manager tiers, and shared workspace context out of the box.

It starts with one decision: every AI interaction begins through one trusted gateway.

Universal AI Gateway

Every provider. One governed funnel.

Connect all 7 major AI providers once, and every module, employee, and workflow calls one internal gateway instead of wiring up its own credentials.

  • One integration point for every provider — business code never touches a provider SDK directly
  • Bring your own encrypted keys, or draw against your company's account balance instead
  • Automatic retry on transient failures, with fallback to a backup model if a provider goes down
  • Real, provider-reported token cost logged on every single request
OpenAIGPT-4o family
AnthropicClaude 3.5
GeminiGoogle's multimodal line
Azure OpenAIEnterprise-managed OpenAI
DeepSeekHigh-efficiency reasoning
GrokxAI's frontier model
OllamaSelf-hosted, on-prem

Single Unified Gateway

AIGatewayService.chat(...) — one call, any provider

Behind that gateway, humans define the routing policy — the platform just carries it out, request by request.

Routing decision

Automatic
Request
OP
AN
GE
AZ
DE
GR
Cost→ routed toDeepSeek
Dynamic Model Switching

Requests route themselves  you set the policy.

Every request is evaluated against live conditions and your configured policy, then routed to the right provider automatically.

Try it — pick a condition and watch the route change. Your policy decides; the platform executes.

The same policies control spending — before AI consumes resources, not after the invoice.

Dynamic Spend Limits

Set the ceiling. We enforce it in real time.

Department, employee, project, and model-level caps — daily and monthly — checked before a request ever reaches a provider.

Quota usage by department

Daily spend — Engineering
62%

$620 / $1,000

Monthly spend — Marketing
84%

$8,400 / $10,000

Per-employee — J. Alvarez
34%

$17 / $50

Project budget — Launch Q3
96%

$960 / $1,000

And because humans set the rules, every AI decision stays transparent and reviewable.

Governed & Auditable AI

Nothing happens off the record.

Every sensitive action across the platform — an AI request, a permission change, a billing event — is captured in an append-only audit trail.

Audit trail

Live-style preview — illustrative events

Append-only
09:41:02M. ChenInstalled module Token Limiter
success
09:38:44SystemFallback triggered claude-3-5-sonnet → claude-3-5-haiku
warning
09:35:17R. OseiApproved waitlist entry acme-labs.io
success
09:31:09A. PatelUpdated role permissions Manager tier
success
09:22:53S. DialloAI request denied Quota exceeded — Sales dept.
denied

Complete audit logs

Every AI request, permission change, and billing event — append-only.

Prompt & usage history

Full conversation and consumption history, reviewable per employee.

Role-based permissions

Custom roles beyond the four built-in tiers, granted per object.

Policy enforcement

Quotas and model restrictions checked before a request is sent.

Approval workflows

Waitlist and provisioning flows route through explicit approval.

Compliance-ready

Structural isolation and encryption designed for audit review.

What gets recorded can be observed — humans continuously measure and improve how their teams use AI.

Real-Time AI Usage Tracking

See exactly where every token goes.

Live dashboards break down usage by model, department, and employee — the same data every request logs automatically.

app.insights.dev/company/analytics

0

Requests today

0ms

Avg. response latency

0%

Success rate

$0

Monthly spend

Request volume — last 24h

Cost by department

  • Engineering38%
  • Marketing24%
  • Support18%
  • Sales12%
  • Ops8%

That feedback loop is where the money is: good governance doesn’t slow innovation — it removes waste.

What Governed AI Unlocks

Governance doesn't slow teams down. It pays for itself.

Illustrative platform-wide benchmarks — a snapshot of what per-employee quotas, automatic fallback, and model routing add up to at scale.

0B+

AI tokens saved

$0

Monthly cost saved

0%

Lower AI costs

0%

Faster AI operations

0+

Teams using the platform

0%

Gateway uptime

Modeled AI spend — unmanaged vs. governed by Sunesix

None of this is bolted on. Here’s the infrastructure that makes trusted AI possible.

Platform Architecture

Every request follows the same governed path.

No side channels, no per-module shortcuts — every layer the request passes through is shared infrastructure.

01Users

Employees and admins across every tenant workspace

02AI Gateway

One funnel — gateway_core mounts every request at /api/v1/

03Governance Layer

Token Limiter and policy modules check quota before routing

04Provider Router

Model registry resolves the right adapter, with fallback

05AI Providers

OpenAI, Anthropic, Gemini, Azure OpenAI, DeepSeek, Grok, Ollama

06Analytics

Real, provider-reported usage rolled up per tenant

07Audit Logs

Append-only record of every sensitive action

08Billing

Plan and usage-based charges against the logged activity

09Reports

Exportable views for admins and platform operators

On that foundation, governance grows as modular capability — installed, not rebuilt.

Module Marketplace

A marketplace, not a monolith.

Every governance capability is an installable module. Turn it on for your workspace — no redeploy, no waiting on a release.

AI Governance

Spend Limiter

Dynamic per-employee and per-group AI spend quota enforcement in dollars, with position-based multipliers and a rolling historical spend ledger.

AI Governance

Agentic Token Reducer

Detects sequential multi-turn conversations stuck in a repeating loop and kills them before further tokens are spent.

AI Governance

MCP Router

Permission-scoped routing to tenant-configured Model Context Protocol servers, governing access to internal filesystems and databases.

AI Governance

File Request Stopper

Blocks or size-caps heavy file uploads (PDF/PPT/DOCX) at the proxy layer, per company or per department.

AI Governance

Image Generation/Upload Blocker

Blocks image generation requests and/or image uploads to the AI Gateway.

AI Governance

Dynamic AI Model Selector

Lets employees or policies pick the best model per request from the model registry.

AI Governance

Auto Clear Context Windows

Automatically clears and resets conversation context after a configurable period.

AI Governance

Optimized Prompt

Rewrites and optimizes prompts before they reach the provider to reduce token spend.

AI Governance

Redundant AI Reruns Stopper

Detects and blocks near-duplicate repeat requests within a short window.

AI Governance

Output Accuracy & Brand Safety

Scans every AI response before it reaches the client — malformed structured output that would crash downstream code, offensive language, and competitor mentions.

AI Governance

Prompt Injection & Security Defense

Blocks and logs prompt-injection and jailbreak attempts before a hostile prompt ever reaches the model or backend systems.

Plus more in active development — see what’s on the roadmap.

Browse full marketplace

And under all of it: trust built into every request, by construction.

Enterprise Security

Built to pass the security review, not just the demo.

Isolation, encryption, and auditability aren't add-ons — they're the architecture every request runs through.

Tenant A

Tenant B

Tenant C

No query path can cross this boundary — by construction, not by convention.

Multi-Tenant Isolation

Query-level isolation via a request-bound context variable — structural, not conventional.

Encrypted API Keys

Every stored provider credential uses field-level Fernet encryption at rest.

Role-Based Access Control

Data-driven roles and permissions, down to individual objects.

Secure Authentication

Dual-portal JWT auth with device/session tracking and forced password resets.

Audit Trails

Every sensitive action — logins, permission changes, AI requests — is logged.

Compliance-Ready

Architecture built around the primitives compliance reviews look for first.

Data Privacy

Employees never see raw provider credentials — decrypted only in-request, in-memory.

Zero Trust Boundaries

No implicit trust between tenants, modules, or provider integrations.

So what does adopting governed AI actually look like? Four steps.

How It Works

Live in four steps.

1

Connect AI providers

Add encrypted credentials for OpenAI, Anthropic, Gemini, or any of the 7 supported providers.

2

Configure policies & limits

Set department, employee, and project token quotas, and pick which modules to enable.

3

Teams start using AI

Employees get access through whatever the Company Admin enables — governed automatically.

4

Monitor & optimize costs

Watch real usage roll in and adjust quotas, models, and modules without a redeploy.

Better yet — experience it yourself, before you sign up.

See It In Action

Explore the platform before you sign up.

A simulated look at the chat, analytics, and module experience your team will actually use.

app.insights.dev/employee
Summarize this quarter’s support tickets by root cause.
Routed to GPT-4o · 214 tokens · within your quota. Three clusters found: onboarding friction (41%), billing confusion (33%), integration errors (26%).

Illustrative preview — not a live session

Now hold that next to the alternative: unmanaged AI, side by side with governed AI.

Why Sunesix

Unmanaged AI access doesn't stay small.

Every one of these problems is invisible on day one, and expensive by month six.

Capability
Unmanaged AI usage
With Sunesix
Provider credentials
Scattered across tools, plaintext in configs
One encrypted vault per tenant, never exposed to employees
Spend visibility
Discovered on the invoice, after the fact
Real-time, per-employee, per-model cost tracking
Access control
Whoever has the API key has unlimited access
Data-driven RBAC, per-department and per-employee quotas
Audit trail
None — no record of who asked what
Append-only audit log on every AI request and policy change
Provider outages
Requests just fail
Automatic retry and fallback to a backup model
Rolling out new capability
New integration, new one-off code
Install a module from the marketplace — zero redeploy

Governance centralized means developers inherit simplicity — one API, one envelope, one schema.

POST /api/v1/ai/chat/200 OK
{  "success": true,  "data": {    "model": "gpt-4o",    "provider": "openai",    "used_fallback": false,    "usage": {      "input_tokens": 214,      "output_tokens": 86,      "cost_cents": 0.21    }  }}
Developer Experience

API-first, because it has to be.

The frontend itself is just a client of the same gateway you'd integrate against — there's no private shortcut.

  • One mount point — every endpoint lives under /api/v1/, nothing scattered across services
  • A live OpenAPI schema at /api/docs/, generated from the real endpoints, not hand-written
  • One consistent response envelope on every call — success and error shapes never vary
  • Official client SDKs are on the roadmap — today, integrate directly against the OpenAPI schema
REST
OpenAPI 3
JWT
View documentation

Enterprise trust isn’t claimed, it’s earned — so here’s exactly what’s true today.

Compliance & Trust

What's true today, stated plainly.

We'd rather tell you exactly what's built than lead with a badge wall.

Encryption at rest

Every stored provider credential is field-level encrypted (Fernet) — plaintext exists only in memory, only during the request that needs it.

Structural tenant isolation

Data isolation is enforced by the query layer itself via a request-bound context variable, not by developers remembering a filter.

Append-only audit logging

Company approval, module publish/remove, waitlist approval, and every AI request are logged — logs are never edited or deleted.

No third-party certifications yet

We haven’t pursued SOC 2 or ISO 27001 yet — the architecture is built around the primitives those audits check first (isolation, encryption, audit trails), but we won’t claim a certification we don’t hold. Talk to us directly about your compliance requirements.

Contact us

And governance doesn’t stand still — it evolves continuously, in the open.

Product Roadmap

What's shipped, and what's next.

Built honestly: modules and integrations below are clearly separated by what's live today versus what's planned.

Shipped

Live today

  • Multi-tenancy, RBAC engine, module registry & marketplace
  • AI Gateway — 7 provider adapters, retry/fallback, streaming, cost tracking
  • Dual-portal JWT auth, audit logging, billing (plans/subscriptions/invoices)
  • 12-module governance catalog, led by the Token Limiter reference module
Next

In active development

  • Real payment gateway integration for billing
  • Second and third full reference modules beyond Token Limiter
  • Push/SMS notification channels
  • Per-tenant AI provider health aggregated into one platform view
Planned

On the roadmap

  • CRM AI, HR AI, Finance AI, Document Intelligence modules
  • Workflow Automation & a preconfigured Employee AI Assistant
  • Slack, Microsoft Teams, Google Workspace, Jira, GitHub, Salesforce, Notion, Zapier, and webhook integrations
  • Official client SDKs on top of the existing OpenAPI schema

Planned modules

CRM AI

Planned

HR AI

Planned

Finance AI

Planned

Document Intelligence

Planned

Workflow Automation

Planned

Employee AI Assistant

Planned

Planned integrations

Slack
Microsoft Teams
Google Workspace
Jira
GitHub
Salesforce
Notion
Zapier
Webhooks

What does all of that add up to? Measurable outcomes — model them with your own numbers.

ROI Calculator

Estimate what governed AI saves you.

A rough model based on the platform-wide benchmark above — plug in your own numbers.

120
$45
Estimated current spend: $5,400/mo

$2,052

Estimated monthly savings

$24,624

Estimated yearly savings

Modeled at a 38% reduction — an illustrative benchmark, not a quote.

The teams already working with us validate the approach.

Built With Early Teams

What design partners are telling us.

Sunesix is pre-launch — these are illustrative quotes from early design-partner conversations, not published customer testimonials.

We went from every team wiring up its own API key to one policy screen. The audit trail alone made our security review painless.

IT

Design partner, IT Director

Mid-market SaaS company

Illustrative

Per-department token quotas meant we could roll out AI access broadly without an open-ended bill at the end of the month.

VP

Design partner, VP Engineering

B2B platform team

Illustrative

Module install/enable with zero redeploys is the part our platform team cares about most — governance changes don't wait on a release.

PL

Design partner, Platform Lead

Internal tools team

Illustrative

Start small, then scale governance as your organization grows.

Pricing

Simple, module-based pricing.

Pay for the modules and AI usage you actually need — nothing bundled you won't use.

Starter

Full platform access for teams bringing their own AI usage budget.

$20 per month

  • All 7 AI providers
  • Every module unlocked
  • $0 bundled balance — top up as you go
Most popular

Growth

Full platform access with a bundled account balance included.

$100 per month

  • All 7 AI providers
  • Every module unlocked
  • $90 of account balance included
  • Priority support

Enterprise

Full platform access, fully usage-based — no base fee.

Custom

  • All 7 AI providers
  • Every module unlocked
  • 100% of cost billed against your account balance
  • Dedicated support & SLA

Need the full comparison? See detailed pricing

A few questions usually remain — answered plainly.

FAQ

Questions, answered plainly.

OpenAI, Anthropic, Gemini, Azure OpenAI, DeepSeek, Grok (xAI), and self-hosted Ollama — all behind one internal gateway abstraction. Connect your own provider credentials, or draw against your company's account balance, funded by the platform's own default credentials, instead.

Ready to govern AI at enterprise scale?

Connect your first provider in minutes, set your policies, and give every team governed AI access — without giving up visibility or control.

Schedule Demo

Get updates as we ship

Join the waitlist for early access and occasional updates on new modules, providers, and governance features. No spam — unsubscribe anytime.

Loading signups… We never share your email