Free IDE extension for risk-free vibe coding — keep secrets out of AI.
Integration · Make.com
Make.com connects AI modules to your business data and workflows at scale. SoterAI Guard integrates via Make's HTTP module to inspect every AI input and output — blocking prompt injection, detecting PII, and preventing unsafe AI responses from reaching downstream apps.
Block prompt injection and jailbreak attempts before data reaches your OpenAI, Anthropic, or custom AI module inside Make.
Surface sensitive identifiers — Aadhaar, PAN, GSTIN, API keys, email addresses — before they enter the AI model context.
Scan AI-generated text for leaked sensitive data, unsafe content, or signs the model was manipulated before results flow to downstream modules.
Log every guard decision with scenario metadata, threat type, and action for compliance reporting and incident investigation.
Create a SoterAI project and get your API key
Sign up at soterai.in, create a project, and copy the API key from project settings.
Add an HTTP module before your AI module
In your Make scenario, add an HTTP > Make a request module before your AI module. Method: POST. URL: https://api.soterai.in/v1/guard/input
Set up the request body and authentication
Body type: Raw. Content type: application/json. Body: { "input": "{{triggerInputField}}", "project": "YOUR_PROJECT_ID" }. Add an Authorization header: Bearer YOUR_API_KEY
Route on the guard response
Add a Router module after the HTTP call. Branch 1: continue if response.action equals allow (pass the redactedInput to your AI module). Branch 2: stop scenario or send alert if action equals block.
Add output guard after your AI module
After your AI module, add another HTTP module calling /v1/guard/output with the AI response. Route on the output guard result to catch leakage in AI-generated text.
Yes. Use Make's HTTP module before and after any AI module (OpenAI, Anthropic, Google AI, or custom). The HTTP module calls SoterAI's REST API to inspect inputs and outputs inline.
Yes. SoterAI is available as a self-hosted Docker deployment. Both Make on-premise and SoterAI can run within the same network, keeping all data inside your infrastructure.
When SoterAI returns action: 'block', you configure Make's error handler or a Router module to branch on that value and stop or reroute the scenario. Make's built-in error handling works naturally with SoterAI's response format.
SoterAI adds under 100ms per guard check. Given that AI API calls in Make scenarios typically take 1-15 seconds, the added latency is negligible.
Add two HTTP modules to any Make scenario and every AI step is guarded. Free tier available.
Get started free