Rota Engine
NewClaude Code support

Every AI call.Under control.

Route.Observe.Tackle.Audit.

One gateway for your models, agents, and developer teams. Govern access, spend, and policies from one place.

Explore the platform
Model providers OpenAI, Anthropic, Gemini, Azure and Mistral connect through the Rota Engine gateway to agents, applications, Claude Code and developer teams.
X:1280  Y:000  V1.0
  1. 01Authenticate
  2. 02Govern
  3. 03Route
  4. 04Trace

Compatible with your model providers
(not an endorsement)

  • OpenAI
  • ANTHROP\C
  • Gemini
  • Azure
  • Mistral AI
  • Jevby
    TypeSafe AI
02Developer teams

Let developers build.
Keep control of the rest.

Bring Claude Code into your gateway. Manage developers, model access, token limits, budgets, and policies by team.

  • Team-level access
  • Developer usage visibility
  • Budgets & policy controls
Explore developer controls
Developer teams
TeamMembersTokens (30d)BudgetPolicy
64.8M24%Standard
42.1M18%Restricted
51.6M16%Standard
Model access
Claude
Token limit
20M per month
Policy
Standard
Illustrative workspaceSelected team: Platform
03

Built for the hard parts of AI.

Five controls. One gateway.
01

Policies that travel with every request

Apply guardrails, filters, and compliance rules across models, teams, and use cases.

Request(user, app, or agent)PII detectionToxic contentData residencyPrompt injectionCustom rulesCleanrequest
02

Spend you can actually govern

Set budgets, allocate spend, and keep costs predictable.

$1,000$500$0JanFebMarAprMayJunBUDGET LIMITAlert before limit
03

Routing that keeps work moving

Automatically route, retry, and fail over across providers.

Request
OpenAIAnthropicGeminiMistral

Routefallbackretrysuccess

04

A trace for every decision

See the full path from request to response, with token usage, policy checks, and more.

  • user request12 ms
  • authenticate28 ms
  • normalize41 ms
  • guardrail check87 ms
  • route to model642 ms
  • model response156 ms
  • audit log12 ms
05

Your stack. Your environment.

Plug into your existing tools, workflows, and run anywhere.

ApplicationsAgentsDataWorkflowsToolsYour systems
  • Cloud
  • Hybrid
  • On-prem
04From request to audit
Nothing out of view.

From request to audit.
Nothing out of view.

  1. Authenticate
  2. Normalize
  3. Guardrail
  4. Budget
  5. Route
  6. Audit
Time (UTC)StageDetailsModelTokensLatencyStatus
12:01:03.000authenticateAPI key validated——12 msOK
12:01:03.012normalizeRequest normalized——18 msOK
12:01:03.030guardrailNo policy violations detected——41 msOK
12:01:03.071budgetWithin team budget——6 msOK
12:01:03.077routeRouted to claude-3-5-sonnetclaude-3-5-sonnet1,284812 msOK
12:01:03.889auditLogged to audit stream——15 msOK
Audit recordIllustrative
{
  "request_id": "req_7f9a3c",
  "model": "claude-3-5-sonnet",
  "tokens": 1284,
  "latency_ms": 904,
  "policy": "standard",
  "status": "ok"
}
05Jev semantic guardrails
RotaEngine with Jev by TypeSafe AI

Read the meaning.
Apply your rules.

Jev-powered semantic checks, enforced through your RotaEngine policies.

Inspect requests and responses for semantic signals. Apply configured actions and keep the result in your request trace.

Explore semantic guardrails
From content to controlJev returns structured decisions. RotaEngine applies the configured action.
Request(user input)Response(model output)Jev semantic checkJev by TypeSafe AIReturns structureddecisions (category,entity, action, etc.)RotaEnginepolicyApplies configuredaction from yourpolicies.Block → StopRequest is stoppedper policy.Warn → Continue + flagRequest continueswith flag in trace.

decision → category: jev:keyword:Medical advice · action: warnIllustrative

Semantic guardrail policies

Policy configuration from RotaEngine.

Both = request + response
PolicyScopeActionDirectionStop on Block
Jev Injection Signature Semantic Guardrail PolicyGlobalBlockBothYes
Jev Keywords Semantic Guardrail PolicyGlobalBlockBothYes
Jev Semantic Guardrail PolicyGlobalWarnBothYes

Flagged. Still traceable.

Source request example

Warn
Sent to model
Category:
jev:keyword:Medical advice
Entity:
Medical advice
Action:
Warn
Provider:
Jev Semantic Guardrail
Type:
flagged

Warn keeps this request moving and records the flag.

  • Semantic checksEvaluate meaning in context.
  • Policy actionsBlock or warn by configuration.
  • Request evidenceInspect category, provider, and outcome.

Input + output checks / Global policies / Traceable outcomes

06Integrations

Your tools, already in the picture.

Connect Rota Engine to your existing stack.

Models

  • OpenAI
  • ANTHROP\C
  • Gemini
  • Jev

Coding

  • Claude Code

Workflows

  • n8n
  • make
  • Temporal

Frameworks

  • LangChain
  • LlamaIndex

Observability

  • OpenTelemetry
  • Grafana
07Get started

Ship AI features fast.
Govern them securely.