API keys everywhere
Provider keys spread across applications and teams, making secure management difficult.
Every AI request.
One gateway.
Scattered API keys, unclear model usage and costs, and separate tool permissions make AI difficult to govern. AI Gateway sits between your agents and LLMs or MCP servers, bringing connectivity, control, and visibility together.
AI agents
Organization & members
- SSO
- Okta sync
- RBAC
AI Gateway
Gateway Key
LLM Gateway
MCP Gateway
Access control
Logs & audit
LLM providers
- Local LLM
MCP servers
- SaaS
- Internal tools
AI operational challenges
As AI usage grows, connections, API keys, and tool permissions become fragmented. Teams need a consistent way to manage usage, costs, and access.
Inconsistent tool permissions
Each agent has its own configuration. It becomes hard to know who can use which tools.
Costs visible only after use
Understanding which team used which resources—and at what cost—takes extra effort.
Model changes mean reconfiguration
Switching models or providers requires reviewing connections across applications.
Unified AI management
Manage model connections, API keys, usage limits, and logs through one gateway. Developers keep their familiar tools while working within your organization’s policies.
Multiple models, one API
Access models from OpenAI, Anthropic, and Gemini through OpenAI- and Anthropic-compatible APIs. Standardize connections across providers and simplify application integration.
Switch models without redeploying
Map the model alias used by your application to an underlying model in the gateway. Update routing without changing the model name in your application.
Secure API key management
Store provider API keys encrypted. Users connect with project-specific Gateway Keys, so you do not need to distribute provider credentials.
Bring MCP tools together
Select tools from multiple MCP servers and expose them through one endpoint. Project policies are checked at execution so only permitted tools can run.
Control usage and costs
Set project limits for requests, tokens, monthly budgets, and request rates. Budgets use estimated costs from pricing tables; models without pricing are marked as unavailable for cost verification.
Unified LLM and MCP logs
Record who used which model or tool and the outcome. Review tokens, estimated costs, and latency in logs and dashboards for daily operations and audits.
Data protection with DLP
Define what to protect and how to respond, then detect and control confidential information in AI inputs. Review findings in audit logs and manage your organization’s data security policies in one place.
Define protection policies
Choose the personal and confidential information to detect and how to respond. Manage your organization’s data security policies in one place.
Detect and control sensitive data
Detect sensitive information in AI inputs and respond according to your policies to reduce the risk of confidential data exposure.
Review results in audit logs
Review findings in audit logs with sensitive values masked. Use the results for follow-up reviews and policy improvements.
| User | Findings | Patterns |
|---|---|---|
| kim.j | Customer personal data 0.96 …Customer contact: j***@example.com… | Phone 3 · Email 2 |
| park.h | HR & payroll 0.93 Content hidden · No pattern match | — |
| lee.s | Confidential documents 0.31 …OPENAI_API_KEY=sk-****… | API key & token 1 |
Security with ACP
Connect QueryPie ACP to protect AI access to internal data and systems. MCP tools provided by ACP give AI secure access to permitted resources.
User
AI models
- Local LLM
AI Gateway
MCP tools
QueryPie ACP
Databases
Servers
Kubernetes
Web applications
Start with AI Gateway
Explore a deployment and operating model that fits your LLM and MCP usage, access policies, and cost management needs.
