The 1-line drop-in
OpenAI replacement.
Add enterprise-grade security, local hardware offloading, and smart cost routing to your existing AI applications — without changing a single line of business logic.
Standard OpenAI Setup
const client = new OpenAI({
apiKey: process.env.OPENAI_API_KEY
});
Spendplane Gateway
const client = new OpenAI({
apiKey: process.env.SPENDPLANE_API_KEY,
baseURL: 'https://api.spendplane.com/v1',
});
No proprietary SDKs.
No Vendor Lock-in.
One of the biggest hurdles in adopting enterprise AI gateways is the requirement to rewrite your existing application stack to use a proprietary SDK. Spendplane fundamentally rejects this approach — our gateway is engineered at the network level to be a 100% compliant drop-in replacement for the standard OpenAI API specification.
Whether your team has spent months building complex agentic workflows in LangChain, RAG pipelines in LlamaIndex, or simple bash scripts wrapped in cURL commands, Spendplane works instantly. You change the baseURL environment variable, and your application natively inherits all configuration, routing, and governance rules. Zero learning curve. Zero technical debt.
-H "Authorization: Bearer sp_key_xxx" \
-d '{"model":"claude-sonnet-4-20250514",...}'
"id": "chatcmpl-gw-9f2a",
"object": "chat.completion",
"choices": [{...}]
}
"Hello, my SSN is ***-**-6789 and my card is 4111-****-****"
"Hello, my SSN is [REDACTED_SSN] and my card is [REDACTED_CC]"
Stop sending sensitive data to the cloud.
Most modern API proxies operate entirely in the cloud, acting as a middleman between your servers and providers like OpenAI or Anthropic. While this provides analytics, it utterly fails the security requirements of healthcare, finance, and legal enterprises. If a developer accidentally logs a patient's Social Security Number in a prompt, that data still traverses the public internet before it can be redacted.
Spendplane's Hybrid Gateway deploys directly within your own secure VPC or local network via our lightweight container. When a prompt is issued, it is evaluated locally. If the prompt contains PHI or PII, it is scrubbed on-device using quantized NLP models before it ever touches an external network. You retain absolute control over your most sensitive data.
Illuminating the shadow-IT black box.
As organizations scale, tracking who is using which models, how much they are spending, and what latency they are experiencing becomes impossible. Individual teams spin up their own API keys, leading to billing nightmares and completely obfuscated API traffic.
The moment traffic flows through Spendplane, the central observability dashboard comes alive. You gain rich, real-time telemetry into every single request. Identify which projects or users are consuming the most tokens. Monitor latency percentiles across providers. Enforce budget limits down to the user level, automatically cutting off access or gracefully downgrading to a local model when spending caps are reached.
Requests
12.4k
↑ 18%
Latency
12ms
p99 overhead
Saved
$842
this week
Deployment in minutes
No complex agents to install on your production servers.
1. Connect Provider
Add your OpenAI, Anthropic, or OpenRouter API keys to the Spendplane Vault.
2. Issue Spendplane Keys
Generate scoped project keys for your developers or automated CI/CD pipelines.
3. Change BaseURL
Update your app's baseURL variable. Instantly get logging, scrubbing, and local routing.
Uncompromising Security Control
Built to support enterprise InfoSec reviews with structured metadata audit logs, pre-egress PII redaction, and region-aware routing.
Audit & Governance Readiness
Spendplane centralizes outbound AI requests and records structured metadata (model, token counts, workspace, latency, and redacted entity types—never raw prompt bodies) so compliance teams can review policy enforcement.
Automated Data Masking
Our proprietary redaction engine combines high-speed Regex patterns with on-device NLP models to identify and obfuscate sensitive information in real-time — SSNs, credit cards, internal IPs, and API keys are caught and masked before leaving your network.
Dynamic Rate Limiting
Protect your infrastructure from billing attacks, runaway scripts, and human error. Configure hard stop thresholds and dynamic rate limiting based on tokens consumed, cost incurred, or request velocity.
Granular RBAC
Integrate with your Identity Provider (Okta, Azure AD) for role-based access. Restrict junior developers to local budget models, grant data scientists frontier access. Generate environment-specific keys instantly revocable with a single click.
Ready to secure your AI pipeline?
Get started in minutes with a free trial, or schedule a guided demo with our enterprise team.